Repositorio Institucional de Documentos

2012 87 Jonathan Correa Magdalena 10B4C Multi-Grid as an alternative to 3He for Large Area Neutron Detectors Departamento Director/es Física de la Materia Condensada Guerard, Bruno Campo Ruiz, Jesús Javier Director/es Tesis Doctoral Autor Repositorio de la Universidad de Zaragoza – Zaguan http://zaguan.unizar.es UNIVERSIDAD DE ZARAGOZA Departamento Director/es Jonathan Correa Magdalena 10B4C MULTI-GRID AS AN ALTERNATIVE TO 3HE FOR LARGE AREA NEUTRON DETECTORS Director/es Física de la Materia Condensada Guerard, Bruno Campo Ruiz, Jesús Javier Tesis Doctoral Autor 2012 Repositorio de la Universidad de Zaragoza – Zaguan http://zaguan.unizar.es UNIVERSIDAD DE ZARAGOZA 10 B 4 C Multi-Grid as an Alternative to 3 He for Large Area Neutron Detectors Colección de Estudios de Física Vol. 31 Esta colección recoge las tesis presentadas en el Departamento de Física de la Materia Condensada de la Universidad de Zaragoza desde su constitución en 1987. Colección de Estudios de Física Vol. 31 10 B 4 C Multi-Grid as an Alternative to 3 He for Large Area Neutron Detectors. Jonathan Correa Magdalena A mi familia Contents Abbreviations xi Introduction 1 1 The Problematic 3 1.1 Introduction............................. 3 1.2 Neutron Scattering Science . . . . . . . . . . . . . . . . . . . . 4 1.2.1 Detector Requirements . . . . . . . . . . . . . . . . . . . 5 1.2.2 Reactor versus Spallation Source . . . . . . . . . . . . . 8 1.3 3 HeCrisis.............................. 11 1.3.1 The 3 He-Counter...................... 11 1.4 Alternative Neutron Detector Technologies . . . . . . . . . . . . 12 1.4.1 10 BF 3 ............................ 13 1.4.2 Scintillators......................... 15 1.4.3 10 B-lineddetectors..................... 16 1.4.4 Summary .......................... 19 1.5 Conclusion.............................. 20 2 Interaction Radiation - Matter 21 2.1 Introduction............................. 21 2.2 Interaction Neutron - Matter . . . . . . . . . . . . . . . . . . . 22 2.2.1 Neutron Capture . . . . . . . . . . . . . . . . . . . . . . 24 2.2.2 Neutron Scattering . . . . . . . . . . . . . . . . . . . . . 25 2.3 Interaction Ion - Matter . . . . . . . . . . . . . . . . . . . . . . 26 2.4 Interaction Photon - Matter . . . . . . . . . . . . . . . . . . . . 28 2.4.1 Photoelectric Absorption . . . . . . . . . . . . . . . . . 31 2.4.2 Compton Scattering . . . . . . . . . . . . . . . . . . . . 32 2.4.3 Electron Interactions . . . . . . . . . . . . . . . . . . . . 32 2.5 Conclusion.............................. 34 3 Basics of Gaseous Neutron Detectors 35 3.1 Introduction............................. 35 viii CONTENTS 3.2 GasIonisation............................ 36 3.2.1 Proportional Counter . . . . . . . . . . . . . . . . . . . 37 3.2.2 Choice of the Filling Gas . . . . . . . . . . . . . . . . . 39 3.3 SignalGeneration.......................... 42 3.3.1 The Electric Field . . . . . . . . . . . . . . . . . . . . . 42 3.3.2 Charge Mobility . . . . . . . . . . . . . . . . . . . . . . 43 3.3.3 Signal Induction . . . . . . . . . . . . . . . . . . . . . . 43 3.4 General Properties of Radiation Detectors . . . . . . . . . . . . 44 3.4.1 Mode of Operation . . . . . . . . . . . . . . . . . . . . . 45 3.4.2 Detector Eciency . . . . . . . . . . . . . . . . . . . . . 47 3.4.3 DeadTime ......................... 49 3.5 Conclusion.............................. 50 4 Detector Simulation 53 4.1 Introduction............................. 53 4.2 Basics of the Monte Carlo (MC) simulation . . . . . . . . . . . 54 4.2.1 Limits of the Simulation . . . . . . . . . . . . . . . . . . 57 4.2.2 The Mono-Layer Approach . . . . . . . . . . . . . . . . 58 4.3 The Boron Converters . . . . . . . . . . . . . . . . . . . . . . . 60 4.3.1 Number of Layers . . . . . . . . . . . . . . . . . . . . . . 60 4.3.2 Neutron Wavelength . . . . . . . . . . . . . . . . . . . . 61 4.3.3 Thickness of the Layer . . . . . . . . . . . . . . . . . . . 62 4.3.4 Pulse Height Spectra . . . . . . . . . . . . . . . . . . . . 64 4.3.5 Threshold Cut-O . . . . . . . . . . . . . . . . . . . . . 65 4.3.6 Other Converters . . . . . . . . . . . . . . . . . . . . . . 66 4.4 DetectorGeometry......................... 69 4.4.1 Geometry of the Voxel . . . . . . . . . . . . . . . . . . . 69 4.4.2 Pressure........................... 70 4.5 Conclusion.............................. 71 5 Multi-Grid Detector 75 5.1 Introduction............................. 75 5.2 Concept............................... 76 5.2.1 DeadZones......................... 77 5.3 Prototypes.............................. 78 5.3.1 MG-12............................ 78 5.3.2 MG-96............................ 79 5.3.3 MG-IN6........................... 81 5.3.4 MG-IN5........................... 82 5.4 Electronics ............................. 84 5.4.1 Front-End Electronics . . . . . . . . . . . . . . . . . . . 84 5.4.2 Readout Channels Reduction . . . . . . . . . . . . . . . 88 Chapter 1 The Problematic 1.1 Introduction Due to the specic properties of thermal neutrons, to be studied in Chapter 2, their detection generally requires a capture interaction with a neutron absorbing element. Secondary particles emitted during this reaction are used for an indirect detection of the incoming neutron. As it will be seen, the list of potential neutron converters is small. Among them 3 He performance has made it the most used converter for neutron detection over the last 30 years. NSS, as it will be seen in Section 1.2, requires very dierent types of detection systems adapted to each instrument. The maintenance of these detectors together with the renovation of present instrumentation demands large quantities of neutron converters. In this context, the need of 3 He claimed by the dierent NSS facilities worldwide for the period 2009 - 2015 is presented in Table 1.1. The present 3 He shortage, to be discussed in Section 1.3, makes the construction of new large arrays of PSD impossible. Instruments more aected by the shortage are those covering large active areas (dened as Large Detectors in Table 1.1): powder diractometers and specially ToF spectrometers. The new ESS, to be built in Sweden, and not yet taken into account in Table 1.1, will require large active areas ≈ 230 m 2 . Three dierent alternative techniques to 3 He have been dened by the scientic community to be developed, evaluated an compared to present detectors: 10 BF 3 -lled gas counters, 6 Li-ZnS(Ag) and 10 B 2 O 3 -ZnS(Ag) scintillators coupled to WSFs and 10 B-lined gas detectors. These techniques are shortly reviewed in Section 1.4. 4 Chapter 1. The Problematic Facility Maintenance Small Detectors Large Detectors ORNL (SNS) 100 1,300 17,100 ORNL (HFIR) 100 1,210 1,060 LANL 100 1,994 12,362 NIST 100 560 - BNL 100 180 - FRM II 50 650 4,500 HZB 100 520 7,850 ILL 100 1,000 3,000 JCNS 40 15 7,200 LLB 50 600 600 PSI 50 - 2,000 ISIS 100 400 11,300 J-PARC 100 40 16,100 JRR-3 31 71 - BNC/KFKI 50 118 500 Total 1,171 8,540 83,572 Table 1.1 : Projected demand of 3 He for detector in the dierent neutron scattering facilities in the period 2009-2015 [1]. 1.2 Neutron Scattering Science NSS is a powerful tool that allows to investigate the structure and dynamics of matter. It is complementary to other types of radiation as X-rays, electrons and ions. Reasons for this rely on the peculiarities of thermal neutrons. They are electrically neutral and have low kinetic energy ( ∼ 25 meV). As a consequence, neutrons can deeply penetrate into matter. Precise information on the magnetic behaviour of materials at atomic level can be collected. Diusive motion in solids and liquids can be studied. Stresses in materials can be investigated, etc. Matter is studied through the analysis of the scattering vector Q=k−k0 , seen in Figure 1.1. k and k' are the wave-vectors of the neutrons before and after the interaction with the atoms that make up the sample. Detectors measure the intensity of the scattered neutrons, I ( 2θ ), as a function of the scattering angle. Information on the momentum transfer, ~Q , during the interaction is measured. Dierent techniques are used according to the type of sample and the information to be obtained. In some cases, energy information can be acquired. A particular case is the ToF spectrometer. The energy transferred during the interaction, ~ω∝ k 2 - 1.2. Neutron Scattering Science 5 Figure 1.1 : Scheme of a NSS experiment. Q is dened as a function of the wave-vectors k and k'. k' 2 , or in terms of the neutron velocity ∝ v 2 - v' 2 . The energy transfer can be then accessed by measuring the delay, ∆ t, between the arrival of the inelastic scattered neutrons compare to the elastic ones. The complete scattering function S (Q, ω ) is then obtained through a transformation from the neutron intensity function, I (2 θ , ∆ t). 1.2.1 Detector Requirements Specications of a neutron detector vary as a function of the type of instrument and the neutron source which feeds it. ToF instruments normally require actives areas of tens of square meters. Typical values for the spatial resolution are in the order of 2 cm × 2 cm. The most important ToF instruments are AMATERAS at J-Parc, CNCS at SNS, LET at ISIS, TOF-TOF at FRM II, NEAT at HZB and IN5 at ILL. All of them have similar requirements and shown comparable capacities. IN5 has been chosen as a model for this study. With an increasing brightness and time structured pulses, new spallation sources are becoming an important part of NSS. Some particularities of these facilities compared to reactor-based ones need to be taken into account for the detector optimisation. Cold Neutron Spectrometer - IN5 IN5 is a high precision direct geometry ToF spectrometer. It is used to study low-energy transfer processes as a function of the momentum transfer [2]. Typical conditions of measurements are about 1% for the energy resolution (e.g. quasi-elastic scattering in solids, liquids, molecular crystals and inelastic scattering with small energy transfers in the range 10 µ eV - 100 meV). An incoming neutron will lose or gain energy in its interaction with the 6 Chapter 1. The Problematic atoms of the sample. This energy transfer  although small  can be measured. Neutron energy is related with its velocity and thus with time. Energy information is extracted from the time delay ∆ t = t - t 0 , of the inelastic scattered neutrons compare to the elastic ones (no energy transferred). Neutron velocities for a wavelength of 1 Å are ≈ 1 km/s. For a typical ToF instrument ∆ t becomes then ≈ 10 µ s. Initial time t 0 , is given by a chopper system shown at the IN5 layout in Figure 1.2. This device allows the energy transfer measurement but decreases the ux. Figure 1.2 : Layout of the IN5 instrument. The ight path of the neutron from the sample-position to the detector is 4 meters long. An active area of 30 m 2 is covered in this instrument. The slit width of the chopper system is the main contribution to the time, and hence energy, uncertainty. To minimise this factor, longer ight times are needed. A large sample-position detector distance is then required. In the case of IN5, this distance is set in 4 meters. This fact, together with a broad angular coverage, explains the large active area required for ToF detectors. In order to minimise the interaction of scattered neutrons before they reach the PSDs, they are placed in a vacuum chamber. The pressure dierence that the detector tubes need to maintain is therefore increased. Thicker walls are needed which implies a higher interaction probability of the neutron within the materials used (stainless-steel in the case of IN5). Lower detection eciency and higher background are the consequences. As a part of the ILL Millennium Programme several major improvements were performed in the instrument. These renovations included the complete substitution of the old 3 He counters for a new PSDs [2]. As it is shown in Figure 1.3, the 3 He-based PSD is located at the back of a 85 m 3 chamber. The made in-house detector by the ILL Detector Group covers a surface 6 times larger than the previous one keeping eciency (80 % @ 5 Å) in similar values [3]. 1.2. Neutron Scattering Science 7 Figure 1.3 : The IN5 PSD can be seen at the back of the 85 m 3 vacuum chamber. It consists in 384  3 meters high  3 He tubes. The PSD is formed by 12 modules each containing 32 cylindrical tubes. Tubes are lled with a 3 He-CF 4 mixture (4.75 + 1.25 bar respectively). The centres of the cylindrical tubes are 2.5 cm apart. Such a dense packing allows to signicantly reduce the dead zones in the detector (2.5 cm between each module). This arrangement yields a total active surface of 30 m 2 . The angular range goes from -12 ◦ to +135 ◦ . Localisation is done as follows. One dimension is obtained from the red tube. The other dimension is measured with a charge division system, to be studied in Section 5.4.2, which gives the position of interaction along the tube length. The obtained pixel size is 2.5 cm × 2.5 cm. A 10 cm of polyethylene shielding envelope together with a 5 mm B 4 C absorber liner gets a at and homogeneous background with a signal-to-noise ratio of 104 for a standard test 1 . A summary of the new IN5 PSD specications can be found in Table 1.2. The main requirements for the detector of a ToF neutron spectrometer can be condensed into the three reviewed now: • To properly match the energy uncertainty produced by the chopper system, long ight paths  several meters long  are needed. This, together 1 Elastic scattering from a 1 mm thick vanadium sample. 8 Chapter 1. The Problematic Characteristic Value Detector Geometry Cylindrical Tubes Sensitive Area [m 2 ] 30 Flight Path [m] 4 Detection Height [m] 3 Scattering Angular Range -12 ◦ to 135 ◦ Vertical Angular Range ± 20.55 ◦ Solid Angle Covered [sr] 1.8 (0.6 π ) Momentum Transfer Range [Å −1 ] 0.2/ λ - 11.8/ λ Spatial Resolution [cm 2 ] 2.5 × 2.5 Angular Resolution 0.37 ◦ Gas Mixture [bars] 4.75 3He + 1.25 CF 4 3He Volume [litres] 3000 Detector Eciency [%] ≈ 80 @ 5 Å Dead Time / Tube [ µ s] ≤ 10 Signal to Noise Ratio 10000/1 Table 1.2 : Some of the most important characteristics of the new PSD installed at IN5. with a broad angular range leads to large active areas: tens of square meters. Spatial resolution required is in the order of 2 cm × 2 cm for both dimensions. • Detection eciencies in the same range that 3 He-based detectors ( ∼ 80% @ λ= 5 Å) are demanded. A high uniformity of this parameter over the total active area of the detector is required. • High signal-to-noise ratio. γ -contamination and neutron background are the two main sources of noise. They need to be fully understood and minimised. 1.2.2 Reactor versus Spallation Source Steady-state neutron sources are based on a ssion reaction [4]. The schematic process is shown in Figure 1.4. During this reaction an average of 2.5 neutrons is generated. Some of these neutrons maintain the process ongoing in a chain reaction. The rest of them are thermalised and guided to the dierent instruments used for NSS. The most important facility is the ILL in Grenoble, where a constant ux of 1.3 · 10 15 cm −2 s −1 neutrons is produced. Also important are the FRM II in Munich (8 · 10 14 cm −2 s −1 ) and the HZB in Berlin (1.2 · 10 14 cm −2 s −1 ) among some others. 1.2. Neutron Scattering Science 9 Figure 1.4 : Diagram of the ssion process used at steady-state sources for neutron production. A heavy nucleus, usually 235 U, is ssioned through the collision of slow neutrons [4]. Figure 1.5 : Scheme of spallation process for neutron production. A linear accelerator is used to increase the kinetic energy of charged particles which impinge the target formed by heavy nuclei. After the reaction these nuclei become unstable and start emitting dierent types of particles. [4] 10 Chapter 1. The Problematic Yet there is a dierent process to produce neutrons: the spallation. Highenergetic particles (e.g. protons) hit a target made of a neutron-rich element. Heavy nuclei reach a highly excited state and start the emission of dierent fragments including neutrons with a broad energy spectrum. After this, the process is similar to what was seen before for reactor-based sources. A spallation reaction diagram can be seen in Figure 1.5. This process is rather fast ( ∼10−15 s), thus the time distribution of spallation neutrons is mainly determined by the time distribution of the driving particle pulse, generally provided by a linear accelerator. Pulses can go from short ( ≈ 10 µ s) to long (a few ms) lengths [5]. ISIS in UK together with J-PARC in Japan and the SNS in the US are the main spallation neutrons sources at present. They are all considered short-pulses sources. With a pulse length of 2.86 ms, and a rate at the peak of the distribution ≈ 30 times higher than at the ILL, the ESS, to be built in Lund (Sweden), will signicantly increase the neutron ux available today [6]. Pulse shape and period are shown for several spallation facilities and compared to the ILL in Figure 1.6. For a ToF instrument however, the presence of a chopper system will limit the ux at the sampe position and thus over the active area of the detector. Figure 1.6 : Pulse shape and period for several neutron spallation sources. Intensities are compared to the ILL reactor. ESS will be the rst long pulse source with 2.86 ms [6]. Together with its brightness, one of the major advantages of spallation sources is the pulsed shape of the neutron ux. It will be used to extract ToF information in all the instrument suite. As seen above, the energy resolution of a ToF instrument is related with the time uncertainty. One of the contributions to this uncertainty, although not the most important, is the time resolution of the detectors. Thinner tubes would help matching this incertitude increasing the energy resolution. A shorter spectrum in wavelength is expected at the ESS. As it will be 1.3. 3 He Crisis 11 shown in Section 2.2.1, this implies a lower detection eciency for the most common used converters. The latest estimation on the required detector active area at ESS is ≈ 230 m2 (most of it belonging to large detector arrays) [6]. As it will be studied in Section 1.3, with the present availability of 3 He the necessary amount of this converter to instrument such a large area ( ≥ 20,000 litres) will not make possible the construction of the ESS. 1.3 3 He Crisis 3 He is a rare isotope of helium with applications in Homeland Security, medicine, industry, and NSS [7, 8, 9, 10]. The main source of 3 He on Earth is the decay of tritium, shown in Equation 1.1, which has an average half-life of τ = 12.32 years. Tritium was used to produce nuclear warheads in the United States and Russia until these programs were stopped after the Cold War. For many years the supply of 3 He exceeded the demand and a substantial stockpile, shown in Figure 1.7, was accumulated. 3H−→ 3He +e−+ν. (1.1) The dramatically drop of the stockpile which causes the present shortage is mainly due to two reasons. On one hand the deployment of 3 He detectors in the US borders in order to prevent the entry of radiological materials (e.g. plutonium) to be used for terrorism. On the other hand, tens of thousands of litres are being used in large science facilities in the US such as the Oak Ridge National Laboratory (ORNL). An allocation of only 8000 litres per year has been announced until 2015 giving priority to applications without alternative to 3 He. This value is to be compared with the needs of the dierent NSS facilities shown in Table 1.1. To compare the capabilities of the dierent alternative techniques chosen by the scientic community are studied now. They are evaluated and compared to the state-of-the-art detector: the 3 He tube, which performance is also reviewed. 1.3.1 The 3 He-Counter Widely used for thermal neutron detection, the 3 He-counter is based on the capture reaction shown in Table 2.2. The eciency of a 3 He-counter is a function of the factor χ [11]: χ=P×d×λ, (1.2) 12 Chapter 1. The Problematic Figure 1.7 : The 3 He stockpile in the US from 1990 to 2010. Since 2004 the 3 He production can no longer meet the demand. This situation provokes the present shortage [7]. where P is the gas pressure [atm], d is the tube diameter [cm] and λ the neutron wavelength [Å]. For a x λ , higher pressures or thicker tubes are needed to reach higher detection eciencies. However, as it was said above, thicker tubes will decrease the time resolution of the detector and hence the energy resolution of the instrument. With respect to the pressure, it can go from 0.25 to 15 bar depending on the required eciency. A typical PHS is shown in Figure 1.8. The full-energy peak is clearly visible at 764 keV. This value corresponds to the total energy of the neutron capture of 3 He to be studied in Table 2.2. When one of the two particles (proton or tritium) reaches the wall of the tube, a part of its energy is not deposit in the gas: this is the wall-eect, further discussed in Section 3.4.2. Lower energy events are accounted giving rise to the distribution shown in Figure 1.8. To decrease th is eect thicker tubes or higher pressures can be used. As it will be seen in Section 3.2.2, a heavier gas is often added to 3 He to increase its stopping power. As it will be discussed in Chapter 2, γ -ray events show a relatively low pulse amplitude. This fact explains the valley found at ∼ 100 keV in Figure 1.8 which is often used for γ -discrimination. An energy threshold can be set at the lowest point of the valley to separate neutron events from γ -rays. Low values of γ -sensitivity ( ≈ 10 −7 ) are usually reached with 3 He-counters. 1.4 Alternative Neutron Detector Technologies The development of alternatives for neutron detection has become a priority for all NSS facilities. The scientic community, organised through the Inter- 1.4. Alternative Neutron Detector Technologies 19 Figure 1.14 : Scheme of a neutron detector formed by straw tubes built by Proportional Technologies [17]. Figure 1.15 : Scheme of the straw detector readout system [18]. A delay line between dierent straws depending on their position within the detector permits to decrease the number of readout channels needed. Although several tests have been made for NSS their main purpose is the development of portal monitors able to full the requirements of Homeland Security [19] in the US. 1.4.4 Summary Main advantages and disadvantages of the mentioned alternatives to 3 He are summarised and can be found in Table 1.3. The eort of the development has been split into the dierent NSS facilities. 20 Chapter 1. The Problematic Advantages Disadvantages Scintillators • Time Resolution • Counting Rate •γ -Discrimination • Detection Eciency 10BF3• Fast Implementation • High Pressure & Voltage •γ -Discrimination • Toxicity & Corrosion • Detection Eciency 10B -lined • Counting Rate •γ -Discrimination • Detection Eciency Table 1.3 : Summary of the advantages and disadvantages of the alternatives mentioned. A validation process of each one of these alternatives needs to be established. The use of one or other technique will depend on several variables such as the specic requirements of the instrument, the type of neutron source feeding the instrument, etc. A compromise between the technique performance and its availability is to be found. In this context, the detector for the NEAT instrument in HZB, initially designed to be 3 He-based, will use a 10 BF 3 solution. In the frame of the CRISP project [20] a collaboration between the ILL and the ESS has been established. The 10 B-lined technology has been chosen for the construction of a new LAND prototype suitable for a ToF instrument: the Multi-Grid detector. 1.5 Conclusion The need of neutron detectors in NSS was introduced. Special attention was paid to a specic scattering technique: ToF, which requires LANDs. Together with large active areas other requirements are demanded: detection eciency, high uniformity, long-term stability, time resolution and high signal-to-noise ratio. As a benchmark, the cold neutron spectrometer IN5 at the ILL was chosen. The type of neutron source feeding the scattering instrument is also important for the detector design. Main dierences between steady-state reactors and the new spallation sources were discussed. Causes and consequences of the present 3 He shortage were discussed. The three alternatives selected by the scientic community were introduced and analysed: 10 BF 3 gas counters, scintillators coupled to WSFs and 10 B-lined proportional counters. The latter was chosen for the development of a new LAND. Chapter 2 Interaction Radiation - Matter 2.1 Introduction Two are the main contributions which decrease the signal to noise ratio in a NSS instrument: the neutron background and the γ -contamination. In the rst case, the cause is mainly the scattering interaction of thermal neutrons with the elements forming the detector itself. This interaction needs to be minimised reason why it is studied in Section 2.2.2. The γ -contribution to the noise in the detector comes from the environment of the instruments in NSS facilities which, in some cases, can be relatively important. γ -insensitive neutron detectors are therefore preferred. The study of the γ response of the Multi-Grid detector in Chapter 7 will require basic concepts on photon - matter interaction. Some notions can be found in Section 2.4. The interaction of two other types of radiation with matter is useful for this study: neutrons and ions. Incoming neutrons do not carry electric charge and therefore they do not interact with matter through the Coulomb force. In particular, thermal neutrons mainly interact with the nuclei of the medium. This interaction is studied in Section 2.2. In the case of the subsequent particles of a neutron capture reaction, energy values are in the MeV range for typical converters. These reaction fragments carry electric charge and their interactions will principally be with electrons of the converter elements as it will be seen in Section 2.3. 22 Chapter 2. Interaction Radiation - Matter 2.2 Interaction Neutron - Matter The neutron is an elementary particle that naturally occurs in the nuclei of the atoms. It is formed by three quarks, two down and one up . However, at the energy range treated in this study, neutrons can be considered as pointlike particles without any internal structure. Its mass has a value of m= 1.68 ·10−27 kg. Free neutrons undergo a beta-decay reaction after an average lifetime of ≈ 881.5 seconds: n−→ p+e−+ν. (2.1) The energy of a neutron E , can be dened as a function of its wavelength λ , through the De Broglie relation [11]: λ=} mv (2.2) where } is the reduced Planck constant, m is the neutron mass and v its velocity. The kinetic energy of the neutron becomes: E=1 2mv2=}2 2mλ2. (2.3) A classication of the neutron according to its energy, wavelength or velocity can be found in Table 2.1. Label Energy [eV] Wavelength [Å] Velocity [km/s] Cold 0.05 - 5 ·10−3 1025 - 4.045 (3.86 - 978) ·10−3 Thermal 0.005 - 0.5 4.045 - 0.405 0.978 - 9.77 Epithermal 0.5 - 1000 0.405 - 0.009 9.77 - 439.6 Intermediate (1 - 100) ·103 (9.05 - 0.905) ·10−3 (4.40 - 43.71) ·102 Hot (0.1 - 10) ·106 (9.05 - 0.905) ·10−4 (4.37 - 43.71) ·103 Table 2.1 : Neutron classication. As a function of its energy  or equivalently wavelength or velocity  neutrons can adopt dierent names. For a ToF instrument such as IN5, studied in Section 1.2.1, the velocity of the neutron is related with the energy resolution of the instrument as it will be studied in Chapter 7. Independently of the incident radiation, the normalised probability for an interaction to take place is dened by its cross section, σ , usually measured in barn (1 b = 10−28 m 2 ). Each radiation will have a number of energy-dependent partial cross sections which correspond to the dierent types of interactions. 2.2. Interaction Neutron - Matter 23 In most cases, the total cross section is dened as the algebraic sum of the partial ones. σtot = Σiσi. (2.4) When a neutron crosses a medium it mainly interacts with the nuclei. Depending on the energy of the incident neutron and the nature of the medium, two main types of interaction are possible: scattering and capture . In the rst case, the neutron is deected by a nucleus. This interaction is described by the neutron scattering cross section, σscat , and it can be split into two dierent processes depending on the energy conservation: elastic and inelastic. During a capture interaction, the incident neutron is absorbed by a nucleus and secondary radiation is emitted. This interaction is described by the neutron capture cross section, σcapt . As seen in Equation 2.4, the total cross section is dened as the sum of the partial cross sections: σtot =σscat +σcapt. (2.5) The macroscopic cross section, Σ , with units of inverse length, is dened as Σ = nσ , where n is the atomic density of the medium. Consequently from Equation 2.5: Σtot = Σscat + Σcapt. (2.6) When a neutron beam, I0 , impinges into a medium it suers an attenuation process that will exponentially decrease the number of neutrons along depth, x : I(x) = I0e−(Σtot x). (2.7) An important parameter to be used by the MC simulation in Chapter 4 is the average distance travelled by a neutron between collisions (or the mean free path, l ). It can be dened as the inverse of the total macroscopic cross section: l=1 Σtot . (2.8) The low energy transfer during thermal neutron scattering ( ≤ meV) prevents this phenomenon to be used for neutron detection. Neutron capture offers access to secondary radiation in the form of charged particles with higher energies ( ∼ MeV) making possible the detection. 24 Chapter 2. Interaction Radiation - Matter 2.2.1 Neutron Capture Preferred elements for neutron detection show a high σcapt together with the subsequent emission  in a short time interval ( ≈ 10 −15 s)  of secondary particles with energies in the MeV range. This secondary radiation, with higher energy that the incident neutron and electrically charged is used for the detection. The elements most common used as converters can be found, together with theirs capture reactions, in Table 2.2. 3 He + n −→ 3 H + p + 0.77 MeV 6 Li + n −→ 3 H + α + 4.79 MeV 10 B + n −→ (94%) 7 Li ∗ + α + γ (0.48 MeV)+ 2.30 MeV (6%) 7 Li + α + 2.79 MeV 157 Gd + n −→ 158 Gd ∗ + e − ( ≤ 0.182 keV) 235 U + n −→ H fis. + L fis. + ∼ 180 MeV + 2.5 n Table 2.2 : The principal neutron converters used for thermal neutron detection are shown. Reaction fragments together with Q-values can also be found. Despite their relatively large σcapt for thermal neutrons ( 2.59 ·105 and 680 barns respectively), 157 Gd and 235 U are not widely used for detection applications. In the case of 157 Gd, the relatively low energy ( ≤ 0.182 keV) of the e− spectra produced by the neutron capture prevents its use. In many cases these low energy particles will not be distinguishable from the γ -background. Some applications have however been found [21]. For the 235 U, energies of the reaction fragments are much higher ( ≈ 180 MeV). However, the high density of the element ( ρU > 19 g/cm 3 ) prevents the reaction fragments emitted deep in the converter to escape the medium. This make dicult to achieve eciencies comparable to other converters. For other applications, such as beam-monitoring, ssion chambers coupled to uranium converters are used [22, 23]. Other elements such as 113 Cd have a large σcapt but secondary radiation emitted is normally formed by a broad range of γ -rays. Some applications however have been reported [24]. The energy-dependence of σcapt for the three rst elements in Table 2.2 can be found in Figure 2.1. A similar behaviour is found for the three of them for a wide energy range. This includes the energies of thermal neutrons used in NSS, already studied in Chapter 1. For E = 25 meV, or alternatively λ= 1.8 Å, the capture cross section for 3He is σ3He capt = 5333 . In the case of 10 B and 6 Li they are found to be σ10B capt = 3835 and σ6Li capt = 940 . This means 71.9% and 17.6% of 3 He respectively. 2.2. Interaction Neutron - Matter 25 σcapt [barns] 0.01 0.1 1 10 100 1,000 1e+04 1e+05 1e+061e+06 Neutron Energy [eV] 0.0001 0.01 1 100 1e+04 1e+06 3He 6Li 10B Figure 2.1 : Energy dependence of the neutron σcapt for the three elements more used in neutron detection [25]. 2.2.2 Neutron Scattering As seen above, the main process used for neutron detection is the neutron capture. However, neutron scattering plays also an important role. For thermal neutrons the three elements seen in Figure 2.1 shown a relatively high ratio σcapt / σscat (e.g., ∼103 for 10 B). This does not apply for other elements present in a neutron detector such as aluminium where scattering dominates over other processes. Due to scattering in aluminium, incident neutrons can be deected in the window and other components of the detector. When the total aluminium thickness is relatively thick, scattering could induce a high neutron background and therefore a reduction in the detector performance. The use of aluminium needs to be minimised. Both, σtot and σcapt of 27Al are shown in Figure 2.2 as a function of the neutron energy. While σ27Al capt increases over a wide range of energy, σ27Al tot becomes at for energies between 0.01 - 104 eV. Note the three orders of magnitude 26 Chapter 2. Interaction Radiation - Matter dierence between σ27Al tot and σ10B capt previously seen in Figure 2.1. σ [barns] 0.0001 0.001 0.01 0.1 1 10 100100 Neutron Energy [eV] 0.0001 0.01 1 100 1e+04 1e+06 σtot σcapt Figure 2.2 : Total and capture cross sections for 27Al as a function of the neutron energy. Note the change in the scale for σ compare with Figure 2.1 [25]. 2.3 Interaction Ion - Matter Ions emitted in the most common neutron capture reactions were shown in Equation 2.2. Interactions between these ions and the nuclei of the converter are possible but rare and therefore the main contribution to the stopping of the fragment comes from the electronic clouds [26]. Due to the large mass dierence between ions and electrons the energy transfer per collision is small. A large number of interactions is needed to stop the fragment. The net interaction can be seen as a continuous slowing down process. As a result of these interactions excitation and ionisation of the medium atoms occur. During the rst process, already mentioned in Section 1.4.2, the incident ion excites electrons of the atoms to a higher energy level. Light is emitted during the de-excitation and it constitutes the base of scintillation detectors. In the second case, positive ions and free electrons 2.3. Interaction Ion - Matter 27 are formed. These secondary particles are prevented to recombine by the application of an electric eld. They are drifted and nally collected, as it will be studied in Section 3.2. Ionisation is the principal mechanism by which charged particles in the MeV range are slowed down in a medium. Therefore it constitutes one of the most important methods in radiation detection. The linear Stopping Power 1 , S, is dened as the dierential energy loss of a charged particle within a medium divided by the corresponding dierential path length: S=−dE dx . (2.9) The classical expression for the stopping power was rstly introduced by Bethe [27] and can be written as: −dE dx =4πe4z2 m0v2nB (2.10) where B≡Z[ln (2m0v2 I)−ln(1−v2 c2)−v2 c2]. (2.11) The quantities v and ze are the velocity and charge of the incident ion. n and Z are respectively, the atomic density and atomic number of the medium. m0 is the electron rest-mass and e the elementary charge. I is known as the average excitation energy of the medium and is normally treated as an experimental parameter. Its value is proportional to Z. For non-relativistic ions the Equation 2.11 becomes: B≈Z ln (2m0v2 I). (2.12) After the approximation, the Bethe Equation can be written as: −dE dx ≈nZz2B(v) v2. (2.13) The term B(v) slowly varies with the velocity of the incident particle. Equation 2.13 would then inversely vary with 1/v2 or the ion energy. This can be understood as follows. The energy transfer depends on the time that the ion remains in the vicinity of the electron which will be longer for slower ions. Consequently, the lower the energy of the ions the larger the stopping power. When the velocity of the incident ions, v , becomes comparable with c , the two last terms in B cannot be neglected and S slowly rises again. 1 Also known as specic energy loss or rate of energy loss . 28 Chapter 2. Interaction Radiation - Matter An analysis can also be done regarding the charge of the incident particle. In Equation 2.13, S increases with z2 . An α -particle ( z= 2 ) has a larger stopping power than a proton of the same energy. Stopping power increases with the charge of the ion. For the medium the dependence varies as nZ . This term represents the electron density. Therefore, high atomic number and dense materials show a large stopping power. Let us study now the range of validity of the Bethe Equation. For a low velocity ion, Equation 2.12 becomes negative for m0v2< I/2 . When the velocity of the ion is lower than (I/m0)1/2 , that can be seen as the orbit velocity of the electrons, the Bethe Equation does not hold. In this situation charge exchange between the ion and the medium becomes important. The ion will pick-up electrons from the medium. Its charge will be reduced along the track which will decrease its stopping power. By the end of the track the ion has captured Z electrons and becomes a neutral atom. A useful way to study the stopping power of a charged particle is to plot it along the track length. This is known as the Bragg Curve . An example of an α -particle travelling through CO2 is shown in Figure 2.3. The initial energy of the particle is E = 2 MeV. During most of the track the charge is equal to two electronic charges. The stopping power increases as predicted by Equation 2.13, 1/E. At x ≈ 10 mm an electron is captured and the curve falls o. A new magnitude called range can be introduced. It is dened as the average distance that ions of certain energy can travel into a medium. In the specic case of Figure 2.3 this value is r= 15.1 mm. Due to the random track that each ion undergoes within the medium, the energy loss is a stochastic process. This results in a spread in energy which can also be seen as a range straggle . 2.4 Interaction Photon - Matter As it will be shown in Chapter 7, γ -sensitivity constitutes an important property of the Multi-Grid detector to be tested. The understanding of this phenomenon requires a brief introduction to the most relevant photon - matter interactions. There are three processes by which γ -ray photon-energy is converted into electron-energy: photoelectric absorption , Compton scattering and pair production . During these processes photons abruptly change their energy and Chapter 3 Basics of Gaseous Neutron Detectors 3.1 Introduction After the neutron capture interaction studied in Chapter 2, the subsequently reaction fragments which have escaped the boron-containing converter reach the stopping gas. Due to the Coulomb interaction, when particles pass through a gas they deposit their energy in the medium and create a certain charge, Q . It is the so-called gas ionisation. This phenomenon constitutes one of the most common methods used for radiation detection. To prevent recombination an electric eld is created within the gas volume. Charges drift from the point where they are produced  along the track of the reaction fragments  towards the electrodes according to their sign. The movement of these charges inside the electric eld induce electric signals used to detect particles. An analysis of how signals are induced in the detector is given in Section 3.3. For special geometries, the electric eld can produce gas amplication. This phenomenon is studied in Section 3.2.1. The most common criteria to select an appropriate stopping gas are reviewed in Section 3.2.2. According to this analysis, two gas species were chosen for the operation of the Multi-Grid detector. Results with each of these gases are studied in Chapter 7. In Section 3.4, some general properties of radiation detectors are reviewed. Two dierent operation modes: current and pulse, are studied. Its implementation is chosen as a function of several parameters such as the expected rate and the type of information to be acquired. Detection eciency is dened. This parameter is used in Chapter 4 to evaluate dierent results obtain by the MC simulation. Some of its most 36 Chapter 3. Basics of Gaseous Neutron Detectors important reducing factors are reviewed. The two dierent behaviours of a radiation detector with respect to the dead time are studied. Expressions to estimate the detector dead time are obtained for both behaviours will be implemented in Chapter 7. 3.2 Gas Ionisation When a charged particle passes through a gas medium it losses its energy creating both excited and ionised molecules along its path. After a molecule is ionised the resulting positive ion and free electron are called an ion pair . Free electrons may have enough energy to create further ions. These high energetic electrons are called delta rays . Since their range is much shorter than the primary particle, in most cases no dierence between ionisation created by the incoming particle or by the delta rays can be measured. The energy needed to ionise a gas molecule  or its ionisation potential, I 0  is some 10-25 eV. However since there is other non-ionising processes in the gas, the average energy lost per ion pair, known as W , (which takes all processes into account) is more often used. Its value is not strongly dependent on the gas species, the type of incident particle or its energy. These parameters are shown for several gases in Table 3.1. Gas Z [a.m.u.] I 0 [eV] W [eV] Range [mm] Ar 18 15.8 26.4 11.2 CO 2 22 13.7 33.0 6.64 N 2 28 15.5 34.8 10.2 O 2 32 12.2 30.8 9.92 CF 4 42 15.9 54 4.32 Xe 54 12.1 22.1 6.15 Table 3.1 : The ionisation potential, I 0 , and the average energy lost W , are shown for several commonly used gases in particle physics. Note the short extent of values for I 0 . Ranges are calculated for an α -particle of E = 2 MeV and gas pressures of 1 atm. [30] Once the charges have been created in the gas, the natural tendency of ions to recombine and form neutral atoms is prevented by the application of an electric eld. This eld causes the electrons and ions to drift towards opposite sides of the detector where the electrodes  biased at a certain applied voltage  are placed. As a function of the applied electric eld several regions of operations can be 3.2. Gas Ionisation 37 dened. In one of them, the proportional region , a multiplication in the number of charges is produced. The total number is however kept proportional to the primary charge. 3.2.1 Proportional Counter Proportional counters rely on the phenomenon of gas multiplication . During the drift towards the electrodes both positive ions and free electrons created by the primary ionisation suer many collisions with neutral molecules. At high values of the applied electric eld electrons may reach kinetic energies higher than the ionisation potential of the gas molecules shown in Table 3.1. The generation of secondary ion pairs is produced. The free electrons created after this secondary ionisation will also be accelerated by the electric eld. They will undergo further collisions with neutral molecules which can produce further ionisation. The gas multiplication takes the form of a cascade also known as the Townsend avalanche . The increment in the number of electrons per unit path length is given by the Townsend equation: dn n=α dx, (3.1) where α is the rst Townsend coecient. When integrating Equation 3.1 for a parallel-plate geometry, as it is described in Equation 3.3, the increment on the electron density can be calculated: M=n(x) n(0) =eα·x, (3.2) where n(x) is the density of electrons after a path length x and M is the multiplication factor or gas gain. The avalanche ends when all free electrons reach the electrode. Heaviar positive ions show a lower mobility. They cannot reach a kinetic energy enough to ionise gas molecules. Gas amplication is only caused by electrons. In order to create an electric eld high enough to start an avalanche an important parameter to adjust is the geometry of the detector. Starting with the simple case of the parallel plates, the electric eld is given by: E=−V d (3.3) where V is the applied voltage and the d the distance between the plates. The negative sign is due to the repulsion between charges of the same polarity. To create high electric elds under this conguration high voltages or small 38 Chapter 3. Basics of Gaseous Neutron Detectors distances between the plates are needed. An example of this are the Resistive Plate Chambers (RPCs) described in [31]. Figure 3.1 : Scheme of a cylindrical shape proportional counter where a is the anode wire radius. The electric eld as a function of the distance from the wire is shown. [30] The use of a cylindrical geometry is shown in Figure 3.1. A wire acting as the anode is placed in the middle of a cylindrical tube which serves as the cathode. The electric eld created at a radius r is given by: E(r) = −V r ln(b/a), (3.4) where a is the wire radius and b is the cathode inner radius. High electric elds can be created by using thin wires: standard radii for wires used in proportional counters are in the order of few tens of µ m. In this case the Townsend coecient varies along the path length, α(x) , so a steeper behaviour than for a parallel-plate geometry is found. The minimum electric eld need for avalanche formation is only reached at distances from the surface of anode comparable to the radius of the wire. Under certain conditions secondary ionisation can be kept proportional to the number of primary ions pairs formed. The total number of ions can be however, multiplied by a factor of many thousands (the previously dened as M ). In fact, gaseous detectors have several regions of operation which are shown in Figure 3.2. At very low values of applied voltage the electric eld cannot prevent recombination of original ions pairs. The collected charge is less that the originally deposited by primary ionisation. As the voltage is raised the ion saturation region is reached. Recombination processes can be considered negligible and all ion pairs created by primary ionisation are collected. Further increases in the electric eld do not yield higher signals because the charge have already been collected. This is the typical operation mode of ion chambers [32]. When the electric eld reaches the minimum value for gas multiplication the collected 3.2. Gas Ionisation 39 Figure 3.2 : The dierent regions of operation of a proportional counter are shown for events deposing two dierent amounts of energy. [26] charges multiply: the pulse amplitude increases. Over some region of the applied electric eld the gas multiplication is linear and collected charge will be proportional to the primary charge created in the gas. This is the region where proportional counters operate. Some non-linearities can be found at the edge of this region due to the space-charge eect . Positive ions created during the avalanche drift a long path (from the avalanche region to the cathode) at a relatively low speed. If the concentration of this ion cloud is high enough the electric eld can be modied. If the voltage is further increased this eect becomes dominant. Multiplication takes places until a certain number of positive ions is created. The minimum electric eld is not reached any more and the avalanche is stopped. The output amplitude have the same value and it no longer depends on the primary charge, or equivalently the energy. This zone is known as the Geiger-Mueller region. Although for some applications detectors working in the ion saturation regions have been reported [], the proportional region is more commonly used. Discrimination of other types of incoming radiation, specially γ -rays, needs of a energy studied of the output signals which is not be possible for a detector working in the Geiger-Mueller region. Single proportional counters are used for many applications in NSS. This system has been chosen for the Multi-Grid detector as it will be shown in Chapter 5. 3.2.2 Choice of the Filling Gas Four features of gaseous detectors have been chosen to show the importance of the lling gas. Parameters such as the average energy lost of the gas molecules 40 Chapter 3. Basics of Gaseous Neutron Detectors or the drift velocity are strongly related with detector characteristics as the signal-to-noise ratio or the rate capability. Signal-to-Noise Ratio In order to properly distinguish events from background noise the factor signalto-noise ratio needs to be maximised. Several parameters of the lling gas can be tuned. • As seen above, the average energy lost, W is dened as the average amount of energy required to ionise a gas molecule. It takes into account all the other non-ionising processes reason why its value is larger than the ionisation potential, I 0 . Its value it is not strongly dependent on the gas species, or the incoming particle type or energy. The lower the W the larger number of primary ion pairs created in the gas for the same incoming particle. • Gases showing a higher M or gas gain generate a larger number of secondary ion pairs during the gas multiplication. This fact implies higher amplitude signals generated. • The electro-negativity is a property of some gases (mainly oxygen and water) to create negative ions by the attachment of a free electron. These negative ions can recombine with positive ions and a net loss in the collected charge is produced. The contamination of the gas mixture with a low percent of the mentioned gases could induce an important reduction of the signals amplitude, as high as 30% in some cases [30]. • The gas amplication threshold is dened as the electric eld value by which the gas multiplication starts. The lower the electric elds the lower the bias voltage applied to the detector. For practical reasons gases with a low threshold value are preferred. Rate Capability In order to evaluated the rate capability of a detector as a function of the gas species the drift velocity is studied. A higher drift velocity  for both free electrons and positive ions  implies a shorter dead time of the detector. The whole signal generation process ends faster and, for the same recovery conditions, the detector is ready for a new interaction. 3.2. Gas Ionisation 41 Long Lifetime Long time operation of the detectors requires both good stability and reduced ageing . As it will be seen below, the use of polyatomic gases (eg. CF 4 ) as quenchers results, for high rates, in the deposit of polymers in the anode. This polymerisation causes the ageing of the detector or a reduction of its lifetime. Spatial Resolution Two are the gas parameters related with the spatial resolution to be studied. • The stopping power, was already studied in Chapter 2. For the same gas pressure, the length of the track within the gas volume will be mainly a function of the electron number of the gas molecule. In this sense, gases formed by heavy molecules will reduce the track length of the reaction fragments and therefore allow a better spatial resolution. • Also important is the electron diusion . If the primary electrons drifting towards the anode spread over a large area a degradation on the spatial resolution would be produced. Another important characteristic for the study of lling gases is the quenching . For some gases, ionisation is not the only energy deposit mode when a charged particle travels through it. Radiative de-excitations are possible. Photons generated during this process give raise to photo-electrons and some distance of the original track. Delayed spurious avalanches are generated and can rapidly start a discharge regime. A way to avoid this eect is adding a small amount of a polyatomic gas or quencher . They show several modes of non-ionising energy dissipation (mainly through vibration and rotation of the molecule) which can absorb the photon-radiation without any radiative emission. Typical compounds used for this purpose are hydrocarbon and alcohol families together with CO 2 or CF 4 among others. The polyatomic gas CF 4 was chosen as a lling gas for the Multi-Grid detector. As seen in Table 3.1, its average energy lost is relatively high. Its scintillation properties have been used for a new kind of gas scintillator neutron detectors [33]. Moreover, as it will be shown in Figure 3.4 is a fast gas. As a drawback, ageing problems have been reported []. A second lling gas, Ar-CO 2 (90-10%) with a better stability and lower gas amplication threshold has also been used and compared to CF 4 . Dierent experiments were performed using both gases. Results are presented in Chapter 7. 42 Chapter 3. Basics of Gaseous Neutron Detectors 3.3 Signal Generation After the ionisation process, the charge drift and the gas amplication the signal generation is now studied. The movement of charged particles inside an electric eld produces a signal induction in the electrodes of the detector. Three are the factors to be studied in order to understand how signals are formed in a proportional counter: the electric eld, the charge mobility and the induction phenomenon. 3.3.1 The Electric Field The electric eld, E, produced inside the detector volume when the bias voltage is applied can be dened through the electric potential, ϕ : E=−∇ϕ. (3.5) The electric eld close to the wire is expected to be similar to what was seen for a cylindrical geometry. As it will be shown in Chapter 5, the geometry of the Multi-Grid proportional counter is based on rectangular tubes (20 mm × 10 mm). Lower eld regions are expected at the corners of the voxel. E was simulated with the Gareld suite [34] and it is shown in Figure 3.3 for the Multi-Grid standard voxel. 0.01 0.1 0.2 0.5 1 2 3 Y - Axis [cm] -0.5 0 0.5 X - Axis [cm] -1 -0.5 0 0.5 1 Figure 3.3 : The electric eld for the Multi-Grid tube was calculated with Gareld [34]. Lines show the shape of the electric eld in kV/cm. As expected E shows high values around the anode wire (which is placed in the middle of the voxel). From there it decreases towards the biased walls. Lower eld zones are found at the corners of the rectangular tubes. Primary charges created in these zones will take a longer time to be collected. This 3.3. Signal Generation 43 increment of the detector collection time, t c  to be dened in Section 3.4.1  might induce loses in the collected charge. It is the so-called ballistic decit. To study this eect, ne scans were performed along the longest side of the voxels. Results will be shown in Section 7.2.3. 3.3.2 Charge Mobility Due to the applied electric eld the generated ion pairs will be forced to move away from the interaction point. With a mobility ∼ 10 3 lower than free electrons positive ions are negligible for this part of the study. The net motion for free electrons is a superposition of random thermal velocity together with the drift induced by the eld. The drift velocity can be dened as: ν−=µ−E P (3.6) where E is the electric eld strength, P the gas pressure and µ− the free electron mobility. The electron drift velocity, ν− , as a function of the electric eld is shown in Figure 3.4 for the two quenching gases used in the Multi-Grid detector at atmospheric pressure. ν− is found to be approximately a factor of ten higher for CF 4 than for Ar-CO 2 (90-10 %) for a large range of electric eld values found within the Multi-Grid voxel. This parameter denes the starting time of the avalanche: after the rst ion pairs have been created, free electrons travel towards the anode at this velocity. For the same electric eld and same track, free electrons generated in CF 4 will reach the high eld region around the wire faster than in the case of Ar-CO 2 . 3.3.3 Signal Induction Due to the short distance from their emission to the wire and their higher mobility free electrons created in the avalanche process will rapidly reach the anode. For this reason their contribution is small ( ∼ 1%). The much heavier positive ions will move towards the cathode at a lower velocity. It is this movement which creates a measurable induction current in both the anode and cathode electrodes. This idea was rst introduced by Shockley and Ramo [35, 36] and can be found in Equation 3.7 in its simpler expression. i=q−→ v·−→ E0 (3.7) where q is the charge of the ion, −→ v its velocity and −→ E0 the weighting electric eld. 44 Chapter 3. Basics of Gaseous Neutron Detectors Drift Velocity [cm/μs] 0 5 10 15 Electric Field [V/cm] 100 1,000 CF4 Ar-CO2 (90-10%) Figure 3.4 : Electron drift velocity ν− , as a function of the electric eld, E, for gases @ 1 bar. Ions created during the avalanche drift towards the cathode which implies that, for the anode wire, −→ v is negative. The vectorial product in Equation 3.7 is then negative. As a result, negative-polarity-induced signals are measured in the anode wire. For the cathode a similar explanation can be given. In this case positive ions drift towards the cathode: −→ v is positive. The vectorial product becomes positive. Positive signals are measured in the cathode electrodes. 3.4 General Properties of Radiation Detectors Three properties to be study in the characterisation of the Multi-Grid in Chapter 7 detector are dened now. • The operation mode : two dierent modes are described. The front-end electronics circuits connected to the detector and the various range of application for each of them are analysed. 3.5. Conclusion 51 The main properties of radiation detectors were studied. The two main operation modes were reviewed: current and pulse mode. Detectors requirements of ToF instruments in NSS: good γ -discrimination and high time resolution, prevents the use of the current mode. The pulse one was therefore chosen. Detection eciency was dened. As it was seen in Chapter 1 is one of the major detector properties used to compare between dierent detector technologies. The denition given is used in Chapter 4 to evaluate the dierent MC simulations. Reducing factors of the detection eciency such as the wall-eect and non-ionising processes were addressed. The distinction between intrinsic and absolute eciency is made. Dead time, related with the count rate capability, was also dened. Expressions to estimate it were obtained for both behaviours: paralysable and non-paralysable. The two behaviours can be found for the Multi-Grid detector depend on the processing electronics used. Chapter 4 Detector Simulation 4.1 Introduction Under perpendicular incidence of a collimated neutron beam, I0 , onto a boroncontaining layer, the number of remaining neutrons at a depth x , from the converter surface follows the attenuation law seen in Equation 2.7. According to that equation, 42 µ m of B 4 C are needed to stop 90% of the incident neutron beam. However, the energy deposit process of the reaction fragments in the converter provoke a self-absorption phenomenon. Neutrons captured deeper than the range of the fragments will not be able to escape the layer and therefore will not be detected. The eciency of a single mono-layer of solid converter is then limited to a few percent as it will be seen in Section 4.3.6. In order to reach high detection performances, the boron-containing converter has to be sliced-o into thin layers. The detection process including gas ionisation and amplication and signal collection (already studied in Chapter 3) needs to be performed between the dierent layers. A MC simulation has been used to optimise the individual thickness of these layers in order to maximise the detection eciency. Parameters such as the neutron wavelength, absorption in the substrate, energy deposit in the gas, the voxel geometry and the gas pressure are studied in this chapter. A compromise between the detection eciency and the number of layers used is needed to minimise the number of readout channels and the detector complexity. Results of the MC simulation will be used to improve the design of the Multi-Grid detector to be studied in Chapter 5. 54 Chapter 4. Detector Simulation 4.2 Basics of the MC simulation The neutron wavelength used for the MC simulation was the same that for the characterisation measurements of the detector, λ = 2.5 Å. The capture cross section becomes then σ10B capt = 5326 barn. For the B 4 C used in the Multi-Grid prototype, which will be further studied in Chapter 6, the 10 B atomic density was found to be n 10B = 1.04 · 10 23 cm −3 , which leads to a macroscopic capture cross section of Σ10B capt = n 10B·σcapt = 5.53 · 10 −2µ m −1 . The capture depth can be written as: x=−1 Σ10B capt ln (nu), (4.1) where the factor nu is generated as a uniformly distributed number ∈ [0,1]. x , calculated per each incident neutron, will follow an exponential attenuation as a function of the layer properties contained in Σ10B capt . For a given set of input parameters (number of layers, individual thickness of the layers, neutron wavelength, etc.) the distance between x and the closest point to the interface converter-gas, d , is found. This is shown in Figure 4.1. Figure 4.1 : Scheme of a capture process in the MC simulation. The available energy of the reaction fragment when escapes the converter is estimated through the distance travelled in the layer, desc . Two more random numbers are generated: θ ( ∈[0, π] ) is the angle subtended by the reaction fragments to the layer plane and Ξ which denes the reaction branch seen in Table 2.2. Once set x and θ , the path of the fragment from the conversion point until it reaches the interface, the escape distance 4.2. Basics of the MC simulation 55 desc , is calculated: desc =d cos(θ) (4.2) In the thermal neutron capture reaction of 10B the two subsequent fragments are emitted in opposite directions. This fact together with the multilayer structure used in the detector provokes that only one fragment per capture can be detected at most. For the studied compound, the stopping power S (dened in Section 2.3) of the four dierent reaction fragments is calculated through the SRIM software [27]. The corresponding Bragg curves can be found in Figure 4.2a. For the same layer composition, S is a function of the ion energy and charge. These parameters lead to dierent distribution for α -particles and lithium ions. Stopping Power [eV/Å] 0 20 40 60 80 Distance [μm] 0 1 2 3 4 (a) Energy [MeV] 0 0.5 1 1.5 Distance [μm] 0 1 2 3 4 Alpha 1 (1.47 MeV) Alpha2 (1.77 MeV) Lithium 1 (0.83 MeV) Lithium 2 (1.01 MeV) (b) Figure 4.2 : (a) Bragg curve: the stopping power, S, as a function of the travelled distance in B 4 C for the four dierent reaction fragments. (b) The available energy of the fragments as a function of the distance travelled in the layer. Maximum distances reached by each ion were dened as range . The available energy of each fragment can be calculated as a function of the distance travelled in the layer and it is shown in Figure 4.2b. The maximum distance that an ion can travel was previously dened as range, r . Similar analyses have been performed for several neutron converters in [38]. During experimental tests, and to discriminate against the electronic noise, an energy threshold is usually applied. For the MC simulation this is equivalent to use shorter eective ranges. Both full and eective ranges, for an 56 Chapter 4. Detector Simulation energy threshold set at 100 keV, of the four reaction fragments can be found in Table 2.3. The choice of the energy threshold will be study in Section 4.3.5. Particle Energy [MeV] r B4C [ µ m] Eective-r B4C [ µ m] α1.47 3.35 3.10 1.78 4.25 3.95 7Li 0.84 1.70 1.35 1.02 1.90 1.55 Table 4.1 : Ranges in the layer are calculated for the four dierent fragments. To apply an energy threshold is equivalent to decrease the ranges. Eective ranges calculated for the threshold set at 100 keV are also shown. The MC simulation uses the escape distance, desc , previously calculated per each capture event and computes the remaining energy at the converter-gas interface using the information shown in Figure 4.2b. The available energy per each fragment is found. As seen in Chapter 2, the principal process by which the energy of the reaction fragments is deposited in the gas is the ionisation. This ion-matter interaction can also be simulated with SRIM. The stopping power of the most energetic fragments of the neutron capture reaction in the gas are shown in Figure 4.3a. Alternatively their available energy as a function of the travelled distance can be found in Figure 4.3b. The much lower density of the gas mixture (CF 4 @ 1 bar) compare to B 4 C, explains the much longer ranges found in this case. For both ions, they are shown in Table 4.2. Particle Energy [MeV] rgas [mm] Eectivergas [mm] α1.78 3.92 3.77 7Li 1.02 2.03 1.55 Table 4.2 : For the most energetic fragments ranges in the gas are shown for CF 4 @ 1 bar. As seen before, to apply an energy threshold is equivalent to use a shorter eective range. Three important parameters can be accessed with the MC simulation: • The Conversion Eciency is dened as the number of capture interactions within the Multi-Grid detector divided by the total number of incident neutrons. It can be directly calculated from the attenuation law. • The Detection Eciency , already studied in Section 3.4, is dened as the number of reaction fragments that deposit in the gas an energy higher than the threshold divided by the total number of incident neutrons. 4.2. Basics of the MC simulation 57 Specific Energy Loss [eV/Å] 0 0.02 0.04 0.06 0.08 Distance [mm] 0 1 2 3 4 (a) Energy [MeV] 0 0.5 1 1.5 Distance [mm] 0 1 2 3 4 Alpha (1.77 MeV) Lithium (1.01 MeV) (b) Figure 4.3 : (a) Bragg curves: S, for the most energetic fragments (lithium 1.01 MeV and α -particle 1.77 MeV) along their track in CF 4 @ 1 bar. (b) The available energy of the fragments as a function of the travelled distance in the gas. • The Escape Eciency is dened as the number of reaction fragments able to escape the converter over the total number of incoming neutrons. 4.2.1 Limits of the Simulation The MC simulation previously described is a useful tool used to optimise several parameters of the Multi-Grid design. However, it is not a complete representation of a real detector. Its limitations need to be known and understood. • 2D Simulation Although the neutron capture process and the subsequently tracks of the reaction fragments are three-dimensional processes, in the MC simulation a simplied geometry was chosen. The track of the fragments is calculated in 2D using the angle θ . • Dead Zones The detection eciency values calculated in almost all the results of the MC simulation are intrinsic [26]: they are estimated at the middle of a voxel. In the case of the wall-eect several points along one axis are studied. Finally, when studying the eect of the gas pressure in the detection eciency (in Figure 4.16) an integration along the length of the voxel is done. • Scattering in Aluminium Due to the low neutron σscat of 10 B compared to σcapt ( σ10B capt / σ10B scat > 103 ), this interaction was ignored. Capture 58 Chapter 4. Detector Simulation and scattering processes occurring in aluminium were however, taken into account. The entrance window and the substrates, used to support the boron coating, constitute a non-negligible contribution to the attenuation within the Multi-Grid detector ( ∼ 6% drop at the maximum of the detection eciency for a 30-layer detector as it will be shown in Section 4.3.3). Neutrons interacting in aluminium, either scattered or captured, are considered to be lost for the eciency calculations. Scattered neutrons however have a chance to be detected elsewhere in the detector increasing the neutron background. • Aluminium Impurities The interactions with other elements present in the aluminium alloy used (AL5083) could be important. High concentration (4 - 4.9%) of manganese together with its relatively high σtot for thermal neutrons (two times larger than σ27Al tot ) might induce a nonnegligible eect. 4.2.2 The Mono-Layer Approach Efficiency 0 0.05 0.1 Thickness [μm] 0 2 4 6 8 10 Back-Scattering Transmission Total Figure 4.4 : Detection eciency for a mono-layer as a function of the layer thickness. Two dierent modes can be dened according to the side of the converter from which reaction fragments escape: transmission and back-scattering. 4.2. Basics of the MC simulation 59 As previously said, if only one single layer is used a relatively low detection eciency is reached. This is illustrated in Figure 4.4. Detection eciencies are shown for a mono-layer detector as a function of its thickness. Usual simulation parameters are used: λ = 2.5 Å and 100 keV energy threshold. Two dierent modes can be dened: transmission or back-scattering [11]. In the transmission mode, the escaping fragment leaves the coating on the opposite side that the incoming neutron, as is shown in Figure 4.5a. The optimisation of the thickness takes into account the number of neutrons converted in the layer too deep for the reaction fragments to escape. An optimum value is found at d ≈ 2.70 µ m. Beyond that value the eciency slowly decreases. A larger number of neutron is captured but reaction fragments are created further from the surface and have therefore a lower escape probability. (a) (b) Figure 4.5 : (a) Scheme of a mono-layer detector in transmission mode. Reaction fragments escape the converter on the opposite side of the incoming neutron. (b) In the back-scattering mode, reaction fragments leave the converter on the same side that incoming neutrons. In the back-scattering mode, the reaction fragment escapes the layer from the same side that the incoming neutron as shown in Figure 4.5b. In this case a saturation region is reached at d ≈ 3.10 µ m and beyond that point it remains stable. This is explain as follows. For a thin layer almost every neutron captured creates a reaction fragment that reaches the stopping gas. The number of capture events is however small. When the thickness is increased more neutrons are converted. Some fragments will not reach the gas because they were created too deep in the converter. If the layer thickness is further increased more neutrons will be converted deeper in the layer but they will not reach the gas. No further increment in the detection eciency is produced. 60 Chapter 4. Detector Simulation 4.3 The Boron Converters A thin lm characterisation, to be discussed in Chapter 6, was performed on the boron layers. Three are the main properties to be used as input parameters for the MC simulation: • The lm mass density, ρfilm = 2.242 g/cm 3 . • The 10 B atomic density n 10B = 1.04 · 10 23 cm −3 . • The atomic concentration of each element present in the lm (shown in Table 6.1). 4.3.1 Number of Layers For the specic lm composition, detection eciency is shown as a function of the number of boron layers in Figure 4.6. dopt is dened as the layer thickness that optimises the detection eciency for a given number of layers. For λ= 2.5 Å, dopt varies from 2.70 µ m for a mono-layer detector in transmission mode to 0.69 µ m for a 100-layer detector. The detection eciency rapidly increases with the number of layers. 30 are needed for a 53.47% eciency. Above 50 layers the increment slows down and a 69.96% eciency is reached for 100 layers. As a compromise between detection eciency and complexity for a practical mechanical construction a design with 30 layers was chosen. The corresponding dopt was found to be ≈ 1.0 µ m. A detection prole for a 30-layer detector (all with the same thickness, 1 µ m) was simulated and it is shown in Figure 4.7a. The exponential attenuation of the neutron ux is found. A possible improvement consists on varying the individual thickness of the coating depending on its position within the detector. This is done as follows. The thickness of the last layer is arbitrary xed in a relatively thick value (10 µ m). Since it is a layer working in the back-scattering mode only, the maximum eciency is reached at 3.10 µ m as seen above. However a thicker value was chosen to stop neutrons not converted in the rest of the detector. Another layer is added before the last one. Its thickness is calculated by iteration until a maximum in the total detection eciency is found. With the thickness of these two layers xed the process is repeated for every layer optimising the eciency at each step. This results in a conguration with thinner layers at the front of the detector and thicker layers at the back. The structure of this detector is shown in Figure 4.7b. The detection prole in this case follows a linear 4.3. The Boron Converters 67 Lithium One of the most common lithium-containing compounds used in neutron detection is the Lithium Fluoride which has a mass density of ρ = 2.635 g/cm 3 , 5 times higher than pure lithium. A 100% isotopic enrichment in 6 Li was considered. The eciency of a multi-layer detector as a function of the number of layers is shown in Figure 4.12. In the same way that for boron, the optimal thickness used to calculate the eciency at each step is also shown. Wavelength used was λ = 2.5 Å. Efficiency 0 0.1 0.2 0.3 0.4 0.5 Optimal Thickness [μm] 0 10 20 30 Number of Layers 0 20 40 60 80 100 Detection Efficiency Optimal Thickness Figure 4.12 : The detection eciency of a lithium-based multi-layer detector is shown. d opt , calculated for each number of layers, is also plotted. Detection eciency shows a similar behaviour than for a boron-based converter. For a 30-layer detector the value was found to be 35.25% which is 34.01% lower than B 4 C. For λ = 2.5 Å, 91 layers are needed to reach a 50% eciency. Optimal values of the individual layer thickness are thicker than in the previous case. 13.12 µ m and 5.0 µ m layer thickness optimise the eciency for a 30 and 100-layer detector respectively. 68 Chapter 4. Detector Simulation Detectors made of thin foils of pure lithium have been already used in a MWPC showing a relatively high eciency [39]. However, problems handing this element prevents its use for LANDs. Uranium In this case, a typical uranium-containing compound was used as the converter: UO 2 . As input parameters a 100% isotopic enrichment and ρ = 10.97 g/cm 3 for the mass density were considered. The two ssion fragments used were Barium (79 MeV) and Krypton (111 MeV). Results can be found in Figure 4.13. Effienciency 0 0.1 0.2 0.3 0.4 Optimal Thickness [μm] 5 6 7 8 9 10 11 Number of Layers 0 20 40 60 80 100 Detection Efficiency Optimal Thickness Figure 4.13 : The detection eciency of a uranium-based multi-layer detector is shown. A lower detection eciency was found in this case. A value of 19.02% was reached for a 30-layer detector. As seen before, increasing the number of layers implies higher eciencies. However only a 37.65% was reached with 100 layers. As it was seen in Section 2.2.1, uranium-based neutron detectors are mainly used for beam-monitoring. The values of the optimal thickness move from 7.18 µ m to 5.62 µ m for 30 and 100 layers respectively. 4.4. Detector Geometry 69 4.4 Detector Geometry The ionisation process in the gas due to the escaping reaction fragments is stopped when fragments collide with one of the walls of the voxel. If the energy deposited in the gas did not reach the threshold value the event will not be recorded. This is known as the wall-eect and strongly depends on two parameters: the geometry of the voxel and the pressure and composition of the gas. 4.4.1 Geometry of the Voxel Properties of the Multi-Grid voxel will be fully discussed in Chapter 5. Dimensions however are given here for a better understanding of this eect. They are 2 cm × 1 cm × 1 cm. The maximum track length of the reaction fragments were shown in Table 4.2. For CF4 @ 1 bar they are approximately 2 and 4 mm respectively. Under these conditions, collisions will only be possible with side-walls and therefore close to the corners. The wall-eect can be estimated by simulating a collimated neutron beam while scanning the X-axis of the voxel. For a 30-layer detector dierent PHS as a function of the distance to the wall are shown in Figure 4.14. The PHS distributions change when the collimated beam approaches the wall. High energy fragments are the rst to collide with the wall due to their longer track. For events produced directly at the corner, the PHS has the same distribution that the one produced in the middle of the cell but just half the number of counts. Only fragments escaping this point towards the centre of the cell can be measured, the rest are lost. Detection eciency is computed along the X-axis and it is shown in Figure 4.15. As seen above, the changes in the PHS start at distances to the wall shorter than the maximal track length. The detection eciency however is not modied until a distance ≈ 1 mm when it starts dropping. At the corner of the voxel it has a value corresponding to half its value at the center of the cell. This can be explained by looking at the PHS in Figure 4.14. The distributions below the applied threshold value (set at 100 keV) do not change for distances further than 1 mm. No neutron events are lost under these conditions. When the beam is closer to the wall than this distance more events of the energy distributions cross the energy threshold lowering the counting which leads to a lower eciency. This eect results in a 2.54% drop in the eciency when integrating along the length of the voxel. A value of 50.93% for the absolute eciency is found. 70 Chapter 4. Detector Simulation Counts 0 5,000 1e+04 Energy [keV] 0 500 1,000 1,500 0 mm 0.6 mm 1.2 mm 1.8 mm 2.4 mm 3.0 mm 3.6 mm Figure 4.14 : PHS for dierent positions of the collimated beam along the X-axis of the voxel. A shift in the distribution towards lower energies can be seen when approaching the wall. At the corner the PHS has the same distribution but half the number of counts than in the middle of the voxel. 4.4.2 Pressure The range of the reaction fragments in the gas linearly scales with the pressure. The use of similar pressures in the detector than in the environment where it is to be installed is preferred. If a lower pressure dierence is to be maintain, thinner walls can be used. This will reduce the neutron scattering background and maximise the detection eciency. As seen in Section 1.2.1, the IN5 chamber is which the PSDs are placed is in vacuum. Lower pressures than atmospheric are therefore more suitable. An analysis of the detection eciency as a function of the gas pressure is presented in Figure 4.16. For low pressures the wall eect, already discussed above, becomes important. Due to the longer tracks of the ssion fragments encounters with all the walls are possible. Low values of detection eciencies are found in this region. There is a value of the gas pressure around 50 mbar from which the eciency 4.5. Conclusion 71 Efficiency 0.3 0.4 0.5 Cell X-axis [mm] 0 1 9 10 Figure 4.15 : The wall-eect decreases by half the eciency at the corners of the cell. The overall eciency is 2.54% lower than the intrinsic when integrated along the X-axis. dramatically raises. This pressure corresponds with a certain track length of the reaction fragments which allow them to deposit an energy in the gas higher than the energy threshold. Beyond this point the detection eciency continues to increase and it reaches the 50.93% @ 1 bar. This value corresponds to the absolute detection eciency already discussed in Section 4.3.1 but integrated along the X-axis of the voxel. 4.5 Conclusion The basics elements of the MC simulation were studied. Special attention was paid to its limitations. Rather than a complete detector simulation it is more an optimisation study of several parameters of the Multi-Grid detector design, specially the optimal layer thickness. A mono-layer approach was rst studied. Maximum eciencies for B 4 C 72 Chapter 4. Detector Simulation Efficiency 0 0.1 0.2 0.3 0.4 0.5 Pressure [bar] 0 0.2 0.4 0.6 0.8 1 Figure 4.16 : The absolute detection eciency as a function of the gas pressure is shown. Decreasing the pressure leads to longer track lengths of the fragments and therefore a more important wall eect. were found to be far from detector requirements discussed in Chapter 1. Typical solid converters in radiation detection were also simulated: Lithium Fluoride and UO 2 . Enriched B 4 C was chosen due to its higher detection performances and other properties to be studied in Chapter 6. A detection eciency higher than 50% can be reached with a 30-layer conguration. The individual thickness that optimises the eciency for this number of layers was found to be ≈ 1.0 µ m. The interaction of neutrons in aluminium has been studied and causes a drop in the detection eciency of ≈ 6% at its maximum. Minimising the use of this element would increase the detector performance. Special attention needs to be paid to the scattering interaction, that represents ∼ 90% of the total interaction, which can provoke and increment in the neutron background within the detector. As a consequence of higher interaction probabilities for longer wavelengths a higher detection eciency was found for high values of λ . Interaction in the 4.5. Conclusion 73 aluminium however starts reducing the eciency when λ reaches 15 Å. The wall-eect was analysed. It depends on the length of the reaction fragment tracks. Therefore, it varies with the gas species and its pressure. For CF 4 @ 1 bar, the absolute eciency was estimated in 50.93% over the length of the Multi-Grid detector voxel. If lower pressures are used longer tracks, and therefore larger wall-eect, is expected. Chapter 5 Multi-Grid Detector 5.1 Introduction Principal requirements of a detector for a ToF instrument in NSS were described in Section 1.2.1. The coverage of tenths of m 2 active areas at relatively high eciencies (80% at 5 Å) is needed. The 3 He shortage, studied in Section 1.3, requires the development of a new technique based on alternatives neutron converters. Results of the MC simulation studied in Chapter 4 have optimised a few important parameters for the design of a neutron detector based on 10 B 4 C converters. The Multi-Grid concept is introduced in Section 5.2. For incoming neutrons of λ = 2.5 Å and using 10 B 4 C as the converter, 30 layers are needed to reach a 53.47% detection eciency. The optimal individual layer thickness was found to be ≈ 1.0 µ m. Dierent detector prototypes were designed and built during this study. Specications of each one of them (size, number of readout channels, etc.) are reviewed in Section 5.3. The front-end electronic system used for the Multi-Grid detector is explained in Section 5.4. A possible improvement of the counting rate capability by reducing the detector dead time is explained. Dierent methods used to reduce the number of readout channels are reviewed. The Centre-ofGravity (CoG) algorithm used to improve the spatial resolution is studied. 76 Chapter 5. Multi-Grid Detector 5.2 Concept The Multi-Grid detector is formed by several layers of rectangular counter tubes with neutron converter coated on the inner side. Counters are made of juxtaposed sections, called grids. The most important advantages of the Multi-Grid design are listed below. • It matches the requirements of ToF instruments: large active areas (tenths of m 2 ) and relatively low spatial resolution (2 cm × 2 cm). • It allows to achieve high coating uniformity inside relatively long ( ≥ 2 m) tubes in a simple way. • The readout method is well adapted to achieve good position and time resolution. The detector is composed of a gas vessel lled with grids electrically insulated one from another and stacked to make rectangular counters [40]. Each grid is made of a frame in which blades coated with a neutron converter lm, are inserted. Both the frame and the blades are made of aluminium which has a low cross section for thermal neutrons as it was shown in Section 2.2. A grid lled with coated blades is shown in Figure 5.1. Figure 5.1 : One grid once lled with 14 coated aluminium blades. The dimensions of the voxel are also presented. 5.3. Prototypes 83 Figure 5.6 : Eight Multi-Grid detector lling a vessel of the same size that the present IN5 PSD module. Almost 60 m 2 of boron coating are needed for one module. Figure 5.7 : A scheme of the Multi-Grid concept replacing the present IN5 PSD. A total of 1728 channels are needed to readout the whole detector. The position resolution in the horizontal dimension (dened as Y in Figure 5.7) can be reduce to 4 cm instead of the 2 cm of the present voxel size grouping two anode row of wires. Moreover a resistive network can be implemented in depth. Two channels are then needed to read out the resistive network formed by all the wires (60) of a Multi-Grid column. For the whole IN5 PSD, this means 2 channels × 8 Multi-Grid columns per module × 12 84 Chapter 5. Multi-Grid Detector modules = 192 channels. These channels are read out by a charge division system. A total of 1728 channels are needed to readout the whole detector which is 2.25 times more than for the present 3 He-based one. Approximately 90% of the channels would use a discrimination system which is simpler than the charge division one. With this readout channel reduction the spatial resolution has become: 4 cm in the horizontal dimension, 2.35 cm in the vertical one and 1 cm in depth. 5.4 Electronics The electronics used in the Multi-Grid detector is now treated. The front-end circuit is studied in Section 5.4.1. Possible optimisation are described. The dierent solutions proposed to reduce the large number of readout channels in the MG-IN6 and MG-IN5 prototypes are studied in Section 5.4.2. . 5.4.1 Front-End Electronics An scheme with the front-end electronics used for the individual read out of the Multi-Grid detector (MG-12 and MG-96) is shown in Figure 5.8. Figure 5.8 : A scheme of the Multi-Grid concept replacing the present IN5 PSD. A total of 1728 channels are needed to readout the whole detector. Signals from the anode wires are decoupled from the HV voltage used to create the gas amplication through a HV capacitor. After that process, they are individually amplied and collected by a Data Acquisition System (DAQ). For the cathode signals the decoupling is not necessary and they are directly amplied. Specications for both anode and cathode ampliers can be found in Table 5.2. The DAQ used depends on the experiment: • It can be a signal digitiser, to study the signal shape. 5.4. Electronics 85 • A Multi-Channel Analyser (MCA), used to produce PHS. • Finally a Multi-Channel Controller (MCC) which is used to create images of the active area by looking at the time coincidences between anode and cathode signals. τ [ µ s] Gain [V/pC] Cathodes 2 32 Anodes 2 9 Table 5.2 : Specications of both anode wires and cathode grids ampliers are shown. As it was in Section 3.4.1, when the pulse mode is used and the time constant of the amplier is larger than the charge collection time of the detector ( τ t c ), the energy deposited in the detector (which is proportional to Q) can be accessed directly reading the amplitude of the generated signals. The energy distribution of the Multi-Grid detector was obtained with a signal digitiser and it is shown in Figure 5.9a. Two are the main reasons to know the energy distribution in a detector: • As previously studied in Section 2.4, the main γ -rejection method is based on the energy dierence between neutron and γ -events. • As it will be shown below, data treatment algorithms such as the CoG need as inputs the total energy deposit in the gas. CoG As said above, the CoG is a data treatment algorithm that allows to increase the spatial resolution of detection systems. This is done by studying the quantity of charge (or equivalently the energy) read in each channel. The spatial position of an event, x p is then dened as: xp=ΣiEi·xi ΣiEi , (5.1) where x i is the each one of the readout channels and E i the energy collected in each channel. As said above, position resolutions beyond the wire pitch in a MWPC can be achieved. As it will be studied in Section 7.2.2, in the case of the Multi-Grid detector the CoG is applied to the cathode channels. An increment of the spatial resolution along the wire dimension is expected. 86 Chapter 5. Multi-Grid Detector Counts 0 200 400 Energy [MeV] 0 0.5 1 1.5 (a) Counts 0 500 1,000 1,500 2,000 TOT [μs] 0 1 2 (b) Time [μs] 0 1 2 3 4 Energy [MeV] 0 0.5 1 1.5 (c) Figure 5.9 : (a) The obtained PHS on an anode wire for a Multi-Grid tube. Distribution of the most common branch ratio of the boron capture can be seen. (b) The two peaks shown in the ToT distribution are also produced by the two most common reaction fragments. (c) The relation between the ToT of the signal and the energy of the event is shown for the typical measurement conditions of the Multi-Grid detector. A typical way to implement the CoG algorithm is by measuring the ToT of the readout signals rather than the signal amplitude. For the Multi-Grid detector the ToT distribution is shown in Figure 5.9b. The relation between the pulse amplitude (or energy deposit in the pulse mode) and the ToT value 5.4. Electronics 87 is well known and, for the Multi-Grid detector is found in Figure 5.9c. As can be seen in Figure 5.9b, the signal length (which can be dened as the mean value of the ToT distribution ≈ 1.84 µ s) is entirely dened by the amplier time constant, τ . As it was seen in Section 3.4.3, τ has a critical role in the dead time calculations. A reduction of its value would decrease dead time and therefore increase the counting rate capability of the Multi-Grid detector. Counting Rate Optimisation As it was studied in Section 3.4.1, the time constant of the front-end circuit working in the pulse mode, τ , needs to be larger than the detector collection time ( τ t c ). However, if it is too large, higher levels of noise will be integrated decreasing the γ -discrimination. Moreover neutrons arriving within the period τ after a previous event will not be counted. An optimisation of this parameter is therefore needed to increase the counting rate performances of the detector. To study this phenomenon, a faster pre-amplier was used ( τ = 2.54 ns). Energy and ToT results can be found in Figures 5.10a and 5.10b respectively. Counts 0 500 Energy [MeV] 500 1,000 1,500 (a) Counts 0 1,000 TOT [ns] 100 150 200 250 300 350 400 (b) Figure 5.10 : (a) Energy spectrum calculated integrating the charge collected in the Multi-Grid detector with a fast pre-amplier ( τ = 2.4 ns). (b) The ToT distribution acquired with the same pre-amplier. As it is shown in Figure 5.10a, the major part of the lithium fragments are lost. The reason is that the signal-to-noise ratio of this measurement was relatively low. The α -fragments however are clearly distinguishable. Conse- 88 Chapter 5. Multi-Grid Detector quently, the ToT distribution shown in Figure 5.10b only shows the longer component. This is however not an issue because the aim of this measurement is to estimate the charge collection time of the detector at nominal voltage, t c . In this sense, t c 400 ns seems a good estimation. If an amplier with a time constant of τ≈ 250 ns was used in the MultiGrid detector, the whole charge will be collected. Moreover, this value is 8 times shorter than the 2 µ s seen in Table 5.2. Faster ampliers are proposed to reduce the dead time of the Multi-Grid detector and increase its counting rate capabilities. 5.4.2 Readout Channels Reduction As explained above, if an individual readout system is installed in prototypes MG-IN6 and MG-IN5 a large number of channels will need to be treated. In the case of MG-IN5, 16 times more channels than the present IN5 PSD are required. In this context a reduction in the number of readout channels is necessary to reduce the complexity of the detector but also to keep this technology aordable. Two dierent methods are used. On one hand to connect several grids of the same vertical position together. On the other hand, a resistive network has been proposed to read either 15 wires in depth of the Multi-Grid detector. The latter method is studied in Section 5.4.2. In the rst case the main eect on the detector performance is the reduction on the Multi-Grid global counting rate capability. If a grid is red somewhere in the same line a coincidence with the wires is done. However, if a second neutron arrives within the amplier integration time τ , to any of the grids connected together it will not be counted. In this sense, the optimisation of τ previously discussed could be applicable to reduce this eect. Charge Division This technique originally consisted in using a resistive wire for the gas amplication. By reading the two ends of the wire, the ratio between the amplitudes in both ends could be used to estimate the point where the avalanche took place along the wire. This can be seen from Equation 5.2: QR QL =x x−L, (5.2) where L is the length of the wire, x the point where the avalanche took place and Q R and Q L are the charge readout on the right and left end of the wire. 5.5. Conclusion 89 In the Multi-Grid detector a dierent approach is taken. Charge division is not implemented to determine the interaction position along the wire length. This is done reading out the grids. Alternatively, the 15 wires in depth (axis Z in Figure 5.7) are connected trough a resistive line by placing a resistor, R, between each wire. Two signals are read out at both ends of the resistive line. In the same way that what it was seen in Equation 5.2, the red wire is estimated through the ratio between charge collected on the left and right side of the line. Alternatively, a charge division readout was applied to the cathode signals. However, due to the capacitive coupling between the grids already discussed in Section 5.2.1, not a proper localisation could be achieved by this method. As a drawback higher applied voltages are needed to properly set an acceptable signal-to-noise ratio which decreases the quality of the energy spectrum. Besides, the electronics needed for the read out are more complicated than the rather simple pulse amplitude threshold used in the rest of the channels. 5.5 Conclusion Detector requirements of a ToF instrument were discussed in Section 1.2.1. Also, the results of the MC simulation were seen in Chapter 4. These two sources were used to develop the Multi-Grid concept. This boron-lined detector is based on several layer of neutron converters ( 10 B 4 C in this case). The coating of this compound is made onto planar blades that are thereafter inserted in aluminium frames forming grids. Grids are stacked and can form relatively large ( ≥ 2 m) rectangular tubes. Several types of Multi-Grid detectors were design and built. Specications of each one of them were studied in this chapter. MG-12 and MG-96 were built and tested. Measurements on detection eciency, γ -discrimination, long-term stability, ageing, wall-eect and cross-talk were performed. MG-IN6 was built after decommissioning MG-96 using the same grids but mounted in a dierent geometry. It will be used in a real ToF instrument. A preliminary study was done to built a detector able to replace an IN5 module. Future results of the MG-IN6 will undoubtedly change many of the design details of this prototype. An important number of problems however can be already address with the mentioned study. Dierent aspects of the front-end electronics were treated. Special attention was paid to the minimisation on the number of readout channels. The aim is to reduce the detector complexity. Several ways to achieve this reduction were studied. How these methods aect the ToF resolution or the counting rate 90 Chapter 5. Multi-Grid Detector capability were studied. Based on the ToT calculus, the CoG algorithm was study as a method to improve the spatial resolution of the detector along the wire dimension. This will be further studied in Section 7.2.2. Chapter 6 Boron Converters 6.1 Introduction The need of boron-containing lms as neutron converters for the Multi-Grid detector has been discussed in Chapter 1. Moreover, a MC simulation was performed in order to optimised a number of important parameters for the design of the detector. Its results, shown in Chapter 4, set the optimal thickness of the 10 B 4 C layer in 1.0 µ m. The choice of 10 B 4 C was made according to the long-term properties (chemical, mechanical and electrical) of this compound compared to pure boron. Dierent deposition techniques were studied. A production method that warranties large scale production capability is to be found. A complete characterisation of the lms is required. A better knowledge of the converter properties will help to understand the exact performance of the Multi-Grid detector. Thin lm characterisation techniques together with neutrography and direct neutron absorption were performed. Values obtained from these analyses, such as mass density, atomic concentration of the dierent elements, etc. were used as inputs for the MC simulation. This software was also use to estimate the incidence on the detection eciency of the dierent contaminants. 6.2 Coating Techniques A maximisation of the 10 B-concentration in the converter is needed for a maximal detection eciency of the Multi-Grid detector. Pure 10 B layers were rst considered. However, several properties of 10 B 4 C make it a more suitable can- 92 Chapter 6. Boron Converters didate: it presents high thermal and chemical stability [42]. In this sense, the long operation time of neutron detectors requires the stability of the coating over time. Moreover, it is easier to handle in a deposition chamber and it has a lower bulk resistivity, which enhance the charge extraction in the detector. The main disadvantage comes from the 20% lower 10 B-concentration of 10 B 4 C compared to pure boron. Despite some works done about the implementations of Chemical Vapour Deposition (CVD) for the B 4 C coating [43], the usual technique used is the magnetron sputtering [44]. Reasons for this rely on the lower impurities concentration  specially hydrogen  achieved with magnetron deposition. A review of the chosen technique for the the neutron converter coating of the Multi-Grid detector is made. 6.2.1 DC Magnetron Sputtering Magnetron sputtering belongs to the wider family of Physical Vapour Deposition (PVD) techniques by which the deposition material is vaporised and condensates on a substrate forming lms. Due to the relatively low pressures ( ∼ 10 −4 Pa) used during the process, chemical reactions are ignored. During the sputtering process atoms are ejected from the sputtering target by bombarding the surface with energetic ions [45]. The scheme of a DC magnetron sputtering machine is shown in Figure 6.1. Figure 6.1 : Scheme of a DC magnetron sputtering machine. The principal components are shown: the substrate, the sputtering target and the magnets [45]. 6.4. Neutron Characterisation Techniques 99 This new density is largely dependent on the atomic mass of the dierent contaminants as can be seen in Figure 6.6. Efficiency 0.5 0.51 0.52 0.53 Contaminant Element B4CH C N O Ar Figure 6.6 : Detection eciency obtained with the MC simulation when adding a 10% contaminant concentration to the studied boron-lm. The rst value corresponds with the 10 B 4 C layer studied above. A detection eciency of  = 53.47% was obtained in Chapter 4. When adding a 10% of a contaminant the detection eciency decreases. This reduction is related with the higher density of the lm. Therefore the higher mass number the higher stopping power, as seen in Section 2.3. For hydrogen a small decrease was measured,  = 53.07%. Carbon, oxygen and nitrogen show similar values due to their similar masses. In the case of a 10% argon contamination the detection eciency would drop to  = 49.97 %. 6.4 Neutron Characterisation Techniques Complementary to thin lm techniques, neutrons can be used for the characterisation of the coatings. No direct information on the thickness or the density of the layers can be accessed. However, it provides a direct measure of the 10B -concentration per unit area. A theoretical calculation of the Σ10B tot is presented. Two types of measurements were performed to experimentally 100 Chapter 6. Boron Converters estimate its value: neutrography and direct neutron absorption. Theoretical Calculation As it was studied in Section 2.2, when a collimated neutron beam crosses a medium its intensity reduction follows the attenuation law. In the case of 10 B 4 C we have: N(d) = N0e−(Σ10B tot ·d) (6.3) where N 0 is the incoming neutron ux onto the boron-containing lm, N(d) is the attenuated ux after crossing the lm and d the lm thickness. The 10 B total macroscopic cross section Σ10B tot =n10B·σ10B tot , is dened as a function of the 10 B atomic density and the 10 B total neutron cross section respectively. Contributions from other elements  specially carbon  have been neglected due to their lower cross sections. Also, since σ10B capt σ10B scat , we can ignore the latter contribution and σ10B tot ≃σ10B capt . The energy dependence for σ10B capt has already been shown in Figure 2.1. In the region of interest for NSS (1 - 10 Å), a linear dependence is found between σcapt and λ . For 10 B this relation can be written as: σ10B capt ≃3835 ·λ[ ◦ A] 1.8barn. (6.4) To determine n 10B , results shown in Table 6.1 are used. Knowing the isotopic masses, M i , and the atomic percentage, P i , of each element, the lm atomic weight can be calculated: Mfilm = ΣiMi·Pi= 10.33 [g/mol]. (6.5) Using this value together with the measured mass density, the lm atomic density, nfilm , can be estimated: nfilm = 2.224 g cm3·1mol 10.33 g·NA 1mol = 1.31 ·1023 cm−3. (6.6) When multiplying nfilm by the 10 B-concentration present in the coating, n10B is obtained: n10B=nfilm ·0.793 = 1.04 ·1023 cm−3. (6.7) With this value, the wavelength-dependent Σ10B capt can be calculated: Σ10B tot =n10B·σ10B capt ≃2.22 ·10−2·λ[ ◦ A] µm−1. (6.8) 6.4. Neutron Characterisation Techniques 101 A direct access to Σ10B tot can be achieved by measuring N0 and N(x) for dierent layer thickness's. From Equation 6.3, it follows: −ln (N(d) N0)= Σ10B tot ·x. (6.9) 6.4.1 Neutrography This method allows the measurement of Σ10B tot of relative large samples (as large as the active area of the PSD used) at once. A complete cartography of the sample is obtained in relatively simple measurements. The Bidim-26 was the 3 He-based multi-wire PSD used for this measurement. Its active area covers 26 cm × 26 cm and has a pixel size of 2 mm × 2 mm. Set-Up This experiment was entirely performed at the CT2 beam-line. In order to uniformly illuminate the entire active area of the Bidim-26 the neutron beam must be widen. After the monochromator the beam is approximately 5 cm × 6 cm. Neutrons can be isotropically scattered using a hydrogen-containing material such as Poly(Methyl Methacrylate) (PMMA). A diverging ux is thus, obtained. Figure 6.7 : The Neutrography set-up. After the low-eciency neutron detector used as a beam-monitor, the scatter is placed inside a B 4 C shielding to minimise the background. The scattered beam impinges the window of the Bidim-26 detector where the blades are placed onto the slits of a cadmium mask. The scatterer is placed at ≈ 1 m from the detector, as can be seen in Figure 6.7. A homogeneous beam impinges the detector window. Blades are placed directly on the window in order to minimise parallax errors due to the 102 Chapter 6. Boron Converters divergence of the beam. A cadmium mask was attached to the window of the detector. It hides any material used to attach the blades to the detector and reduces the background. A calibration measurement was done in which the non-coated substrates were placed onto the slits of the cadmium mask. This is equivalent to measure N 0 . Due to the small size of the pixel in the Bidim-26 detector a long exposition time (2 days) is needed to reach good statistics. A second measurement is performed after replacing the aluminium blades for the coated ones. N(d) is then measured per each pixel. The resulting image of this measurement is shown in Figure 6.8. Y - Axis 120 100 80 60 40 20 Counts 5e+05 1e+06 1.5e+06 X - Axis 20 40 60 80 100 120 Figure 6.8 : An image of the Bidim-26 detector used for neutrography. The 16 slits of the cadmium mask are visible. Boron coated blades were placed onto them. Due to the divergence of the scattered beam the edges of the slits show a halo. Information coming from these areas could induce to an error. Regionof-Interests (ROIs) are created within each slit. Σ10B tot is calculated in these areas. Its value, as a function of the position of the blade in the deposition chamber is shown in Figure 6.9. 6.4. Neutron Characterisation Techniques 103 - ln (N(x)/N0) 0.05 0.06 0.07 0.08 Position in the Deposition Chamber [mm] -200 -100 0 100 200 Run 8 / Blades A - B Run 8 / Blades C - D Run 17 / Blades A - B Run 17 / Blades C - D Run 1 / Blades A - B Run 1 / Blades C - D Calculation (1.8 Å) Figure 6.9 : Σ10B tot is plotted for several boron-coated blades as a function of the position in the deposition chamber. A lower absorption than expected was discovered. Also, an important asymmetry was found compare to previous results. Results The use of the scatterer to get a widen beam provokes a thermalisation of the incoming neutrons. After a number of collisions with the hydrogen atoms of the PMMA their energy is modied. In this case, they gain some energy (or equivalently reduce their wavelength, λ ). In Figure 6.9, the theoretical calculation for λ= 1.8 Å, the corresponding λ for room temperature neutrons instead of λ= 2.5 Å, is shown. Blades from three dierent runs  1st, 8th and 17th  were measured. For each of them, blades coated on the dierent regions were studied. The results for the lower part of the chamber (negative values of the X-axis in Figure 6.9) are consistent with what was previously seen in Section 6.3.3. The outer blades, corresponding to region A, shows a gradient in absorption related with the thickness gradient. In the centre of the chamber, represented by blades from region B, the behaviour is more stable. An important asymmetry was found in the upper part of the deposition 104 Chapter 6. Boron Converters chamber (positive values of the X-axis). In this case values for the outer, corresponding with region D, and centred blades, region C, show a lower absorption than the lower half of the deposition chamber. This behaviour was not observed in the SEM measurement. A non completely homogeneous scattered beam could explain the nonasymmetry found in Figure 6.9. This inhomogeneity could also aect the wavelength of the incoming neutrons which is function of the number of collisions that they undergo within the scatter. Present results of neutrography can only be used to analyse uniformity within each deposition region. No reliable information on Σ10B tot can be extracted from it. Improvements in the set-up need to be accomplished to warranty a homogeneous  in both, size and wavelength  neutron beam. 6.4.2 Direct Neutron Absorption Due to the problems observed with the Neutrography a simpler measurement was performed. Instead of a PSD covering a relatively large area a 3 He-counter was used. A wide beam is not longer needed and a collimated neutron beam can be used. Set-Up Figure 6.10 : Scheme of the Direct Neutron Absorption measurement. A collimated neutron beam impinges the coated blades. Transmitted neutrons are detected with a standard 3 He-tube. Instruments CT2 ( λ = 2.5 Å) and T3 ( λ = 7.5 Å) were used. For this measurement the T3 instrument ( λ = 7.5 Å) and the CT2 beam- 6.4. Neutron Characterisation Techniques 105 line ( λ = 2.5 Å) at the ILL were used. An scheme of the standard set-up is shown in Figure 6.10. A collimated monochromatic neutron beam (0.5 mm × 2 mm) impinges the sample. Transmitted neutrons are measured with a 3 He-based counter. Background due to scattering processes within the blades is minimised with the neutron counter shielding. Transmitted neutrons for a blank blade, N 0 = N(d = 0), and several boron thickness's, N(d), were measured. Following Equation 6.9, - ln (N(d)/N 0 ) versus the thickness of the boron coatings is plotted. The trend-line corresponds to Σ10B tot . Results can be found in Figure 6.11. Double-side coated blades were used for this experiment. Results - ln (N(x)/N0) 0 0.2 0.4 0.6 0.8 1 1.2 10B4C Thickness [μm] 0 1 2 3 4 5 6 Data @ CT2 (2.5 Å) Data @ T3 (7.5 Å) Experimental Fit Calculation Figure 6.11 : Σ10B tot is plotted as a function of the lm thickness for several doubleside coated blades. A linear behaviour was found for both wavelengths. As expected, a linear behaviour for Σ10B tot was found for both wavelengths. The last point, corresponding with a double-side boron layer of 6 µ m, was found to be more absorbing than foreseen in both instruments. Such a response could be explained by a slightly thicker coating on this sample. This point 106 Chapter 6. Boron Converters was therefore not used for the trend-line calculations. Lower values for the absorption were found in both cases. For λ = 7.5 Å the disagreement was found to be 4.32%. In the case of λ = 2.5 Å a larger discrepancy 17.58% was measured. A possible explanation for the four times lower absorption found in CT2 compare to T3 could come from a non-entirely monochromatic neutron beam. It is possible that an important percentage of the neutron distribution coming out of the graphite monochromator has a lower wavelength than 2.5 Å. The only possible value of this harmonic contamination is λ = 1.25 Å (3rd harmonic cannot be transported by the neutron guide). This would imply the measurement of a lower Σ10B tot as it is the case. A 30% contamination of this component would be necessary to match the results. Uniformity A variation in the neutron absorption is expected due to the non-uniformity of the coating thickness seen in Section 6.3.1. Direct neutron absorption is now used to conrm this result. Several points were measured along the blade length. Any variation in the coating thickness will yield a uctuation in the transmission prole. Scans were performed with a pitch of 1.2 mm for the four dierent types of blades according to their position in the deposition chamber. Results can be found in Figure 6.12. The gradient in thickness previously discussed can be seen. For coating regions in the outer part of the deposition chamber, a lower absorption close to the edges is found. Blades placed in central regions show a more stable response. The same behaviour is found at CT2 with λ= 2.5 Å. To achieve a high uniformity over large areas for the Multi-Grid detector this phenomenon has to be taken into account during the mounting of the blades. A specic mounting conceived to improve the uniformity will be explained and discussed in Section 7.4. 6.5 Conclusion As a boron-containing compound 10 B 4 C was chosen over pure 10 B due to its higher reliability (thermal, chemical, etc.) which makes easier the deposition process. Long-term stability, high lm quality, low impurities and high 10 B concentration, were achieved with the chosen deposition technique: DC magnetron sputtering. Layer thickness's were measured using SEM. With this parameter known, 6.5. Conclusion 107 - ln (N(x)/N0) 0 0.1 0.2 0.3 0.4 Position in the Deposition Chamber [mm] -200 -100 0 100 200 Blade A Blade B Blade C Blade D Calculation (2.5 Å) Calculation (7.5 Å) Figure 6.12 : Σ10B tot measured at dierent positions corresponding to the four different regions in the deposition chamber: A, B, C and D. Results were obtained at two dierent instruments with two dierent wavelengths. the lm density was estimated by the use of XRR. Its value was found to be ρB4C= 2.24 g/cm 3 . Film isotopic composition was determined as a function of depth with ERDA. MC simulation were performed to estimated the reduction in the eciency due to the presence of contaminants in the lms. Neutrography and direct neutron absorption measurements were performed in two dierent instruments (T3 @ λ= 7.5 Å and CT2 @ λ= 2.5 Å) were used to validate the previous results. Σ10B tot was measured as a function of the layer thickness. As expected, an linear behaviour was found. Experimental values were found to be close to theoretical expectations at T3. At CT2 however, a lower absorption than predicted was measured. The idea of non monochromatic neutrons at this beam-line has been proposed as an explanation. A 30 % 1st harmonic contamination produced in the monochromator would be enough to match the experimental results. A thickness gradient within the deposition chamber was measured with SEM and conrmed with the Direct Neutron Absorption measurement. Four dierent types of blades were dened according to the coating prole. The 108 Chapter 6. Boron Converters outer sections of the chamber showed a higher gradient (13 % along the blade length) than the central regions (only a 5 %). This information is taken into account for the Multi-Grid detector mounting as it will be explained in Section 7.4. Neutrography shows a non-asymmetry not detected with SEM or Direct Neutron Absorption. Also, values found for Σ10B tot are lower than expected. A reason could be the non-homogeneity of the scattered neutron beam impinging the detector. Improvements in the set-up should be implemented to correct these eects probably coming from a non-homogeneous (neither in shape nor in wavelength) scattered beam. 7.2. General Characterisation 115 Efficiency 0 0.1 0.2 0.3 0.4 0.5 Position [mm] 0 5 10 15 20 Grid 4 Grid 5 Sum G4 + G5 Figure 7.6 : Detection eciency was measured simultaneously for targeted and neighbour grids at every point of the Scan 1. Values in the middle of the studied voxels are maximal and similar in both grids. However when travelling from the centre of the cell towards the other grid a drop in eciency is produced. It starts at approximately 4 mm from the edge. This distance corresponds with the maximum range of the α -particle travelling in CF4 @ 1 bar (as seen in Figure 4.3b). The sum of both eciencies is also shown in Figure 7.6. When the collimated beam is far from the edge its value does not diers from the value of the targeted grid. However, when the beam targets positions close to the edge the cross-talk between grids attenuates the drop in the detection eciency and the value for the sum only decreases by ∼ 5%. Spatial Resolution Improvement After what it was seen in Figure 7.5 the idea of using this cross-talk between grids as an advantage of the Multi-Grid geometry came up. In Figure 7.7, the amplitude measured on the targeted grid plus rst and second neighbouring grids is plotted versus wire amplitude. 116 Chapter 7. Detector Characterisation Amplitude on the Grid [V] 0 1 2 3 Amplitude on Wire 1 [V] 0 0.5 1 Targeted Grid 1st Neigh. Grid 2nd Neigh. Grid Figure 7.7 : Maximum amplitude measured on the targeted and 1st and 2nd grids versus maximum amplitude on the Wire 1. This signal multiplicity for a single event seen in Figure 7.7, can be used to improve the Multi-Grid spatial resolution along the wire dimension. As it was studied in Section 1.2.1, the pixel size of IN5 is 2.5 cm × 2.5 cm which is already slightly larger than the Multi-Grid (2 cm × 2 cm). This improvement however could lead to the construction of thicker grids (and therefore wider aluminium blades) and therefore a reduction in the number of grids to be used to cover the same active area. Using a simple localisation algorithm, based on the center of gravity between dierent frames, some preliminary results were found and are shown in Figure 7.8. For the two points plotted in Figure 7.8 and improvement in the resolution is obtained. In the case of the position x = 4 mm, although the broad distribution found, the estimated Full Width Half Maximum (FWHM) was found to be 1.01 cm. This is half of the detector spatial resolution (which is dened by the cell dimension for an individual readout). When the beam targets the edge of the cell at x = 10 mm, a greater improvement in the spatial resolution is obtained: it was found to be 0.3892. This is a factor of 5 better resolution that before. 7.2. General Characterisation 117 Counts 0 200 400 600 Position Along the Wire [cm] -2 -1 0 1 2 Actual Beam Position Estimation at x = 4 mm Estimation at x = 10 mm Figure 7.8 : Two dierent distributions estimated for x = 4 and 10 mm from the center of the cell respectively. 7.2.3 Wall-Eect When scanning along the same grid a dierent phenomenon can be studied: the wall-eect. The path of Scan 2 was shown in Figure 7.3. The targeted grid was G4. PHS measured while moving from the targeted row (R3) to the neighbouring one (R4) are shown in Figures 7.9a and 7.9b respectively. The PHS obtained for the three rst points at the targeted row (x = 0, 2 and 4) are similar: no eect of the wall can be seen. However, when the collimated beam moves further towards the corner, higher amplitude signals start getting lost at x = 6. This eect increases for positions closer to the wall. At the edge, x = 10, the MC simulation shown in Figure 4.14 estimated a PHS corresponding to approximately half of the one obtained at the middle of the cell. This eect is seen in Figure 7.9a for the lithium fragments which show amplitudes at the border which are ≈ half of the ones at the middle of the cell. For the more energetic α -particles however the ratio between this two positions seems to be closer to 30% than to 50%. Neighbouring grid shows the same PHS than the targeted one at the corner 118 Chapter 7. Detector Characterisation Counts 0 5,000 1e+04 Energy [keV] 0 500 1,000 1,500 (a) Counts 0 5,000 1e+04 Energy [keV] 0 500 1,000 1,500 x = 0 mm x = 2 mm x = 4 mm x = 6 mm x = 8 mm x = 9 mm x = 10 mm (b) Figure 7.9 : (a) PHS obtained at the targeted row for dierent points of Scan 2. (b) For the same points, the PHS measured in the neighbouring row. of the cell (x = 10). The events produce here are split into both rows due to the size of the neutron beam. This also applies for x = 9 which shows a higher counting than other positions. The rest of the measured positions show a measurable PHS due to the fact that the wire is collecting signals all along its length which covers 12 grids. This eect is completely suppressed for the detection measurement studied below which used a coincidence set-up. In the same way that for the cross-talk, the detection eciency is measured per each point of the Scan 2. Results can be found in Figure 7.10. Detection eciency at the center of the cell shows a relatively uniform distribution. It starts dropping at approximately 2 mm from the wall. At x = 10 the value for the sum of eciencies is minimal ( ∼ 33 %). This result is in relatively good agreement with the MC simulation studied in Section 4.4. The drop in detection eciency was however expected to start at positions closer to the wall. In this case and due to the size of the neutron beam the wall-eect can be seen from more than 2 mm apart. Another eect of the relatively large size of the beam is the lower drop in the eciency compared with the expected value. 7.3. Eciency 119 Efficiency 0 0.1 0.2 0.3 0.4 0.5 Position [mm] 0 5 10 15 20 Row 3 Row 4 Sum R3 + R4 Figure 7.10 : Detection eciency measured simultaneously for the targeted and neighbouring row along the points of Scan 2. 7.3 Eciency Experiments took place at CT1 and CT2 beam-lines. Three X-Y slits were used to collimate the neutron beam which impinges the detectors to a size of 2 mm × 2 mm. A beam-monitor was placed before the slits to control the stability of the neutron ux provided by the nuclear reactor. 7.3.1 Reference Measurement The 3 He-based detector shown in Figure 7.11a was used to estimate the neutron ux for the eciency measurements. Its internal structure consists in 37 hexagonal tubes (8 mm diameter). The gas mixture of the detector was set in 3.5 bar of 3 He and 1.25 bar of CF 4 acting as a gas quencher. Measurements are done by targeting the center of the middle row of tubes. When plotting the cumulative counts as a function of the number of tubes in depth (i.e. along the z axis), the behaviour shown in Figure 7.11b is found. By applying a tting curve to the data, an experimental estimation of the total 120 Chapter 7. Detector Characterisation (a) Cumulated Counts 0 2,000 4,000 No. Tubes 0246810 (b) Figure 7.11 : (a) The 3 He-based reference detector. The 37 hexagonal tubes, 8 mm diameter can be seen. The neutron beam impinges on the central row of tubes. (b) Flux estimation - the seven points are cumulative counts in depth for the reference detector. The exponential t gives an experimental estimation of the total ux. neutron ux, Fest , can be determined. The tting curve is dened as: y=a·(1 −e−(x·b)), (7.1) where a = 4.62 · 10 3 and b = 0.3554, are the tting parameters. The ux for the rst tube, the most exposed one, is 1387 neutrons/s (over the size of the collimated neutron beam). The eect of the dead time in this detector reduces the neutron ux by less than 0.1%. 7.3.2 Multi-Grid detector The same set-up as described previously is used. The reference detector is removed. Neutrons now impinge on the middle of a voxel row of the MultiGrid prototype lled with 1 bar of CF 4 . B 4 C shielding is used to reduce the neutron background. Thresholds are set above the electronic noise level. Events are recorded when two signals, one from the red grid and the other from any of the 15 wires, are measured in coincidence. Total counting per wire, N wire , is then compared to the total ux estimated by the reference detector, 7.3. Eciency 121 F est , over the same acquisition time. εwire =Nwire/Fest. (7.2) Eciency per wire can be determined and is shown in Figure 7.12. The exponential attenuation of the neutron beam in the detector is observed. Efficiency 0 0.02 0.04 0.06 Cells 2 4 6 8 10 12 14 MC Simulation Data @ CT1 Figure 7.12 : The eciency per cell in depth. The exponential attenuation of the neutron ux is clearly visible. As expected, the rst and last tubes, with only one boron layer, show a lower eciency than the rest. The overall eciency is calculated as the sum of individual wire eciencies. The overall eciency is obtained as the sum of each wire: εn= Σ15 1εwire . A value of εn = 53.09% ± 0.77% (stat.) was found. This result is to be compared with the 53.47% eciency found for the MC simulation shown in Figure 4.6. Although the maximum eciency is found to be in good agreement with the expected value from the MC simulation the shape of both exponential attenuation show an important discrepancy. Possible reasons for this are discussed now. 122 Chapter 7. Detector Characterisation 7.3.3 Layer Thickness As shown in Chapter 5, grids using six dierent layer thickness were used for the MG-12 detector. The experimental results are shown in Figure 7.13 together with the simulated curve seen in Figure 4.9. When plotting the detection eciency versus the layer thickness the maximum of the distribution is found around 1 µ m. Efficiency 0 0.1 0.2 0.3 0.4 0.5 Layer Thickness [μm] 0 2 4 6 8 10 MC Simulation Data @ CT1 Figure 7.13 : Detection eciency as a function of the individual layer thickness for the MG-12. Five dierent coating thickness were used. For comparison, the MC simulation shown in Figure 4.9 is replotted here. A discrepancy was found between the measured values and the MC simulation. Possible explanations are studied. 7.4. Uniformity 123 7.4 Uniformity The variation of coating thickness among the various blades used is measured using Si wafers mounted in dierent positions on the sample holder within the sputtering chamber. A 13% gradient in the coating thickness along the length of the blade was measured using thin-lm characterization techniques, as seen in Section 6.3.1. 7.4.1 Variation measurement Based on the results shown in Figure 6.4, several dierent types of frames were assembled based on the maximum gradient expected in the blades used. In some frames, blade orientation was alternated, so that for each detector cell, the thickness (and, therefore, eciency) varies in opposite directions in the front and rear coating. This largely cancels the overall eect on eciency across that frame. In a number of the frames, however, this compensation was deliberately not implemented. Also the blades with the greater gradient were used in these frames. Therefore a 13% variation in total amount of 10 B from one side of these frames to the other is expected. MC simulations show that varying the coating thickness by up to 13% around its optimal value (1.0 µ m) yields an eciency variation of ∼ ± 1%. To verify this eect, a uniform scan of the detector was performed. An Am-Be neutron source was translated along the detector at a constant rate over a long time (several days). The number of counts in each voxel was recorded as a function of time. Each grid is read out by an individual amplier, and a signal recorded when coincident with a wire signal. A constant threshold was used in all ampliers and therefore grid-to-grid count variation is dominated by the eect of the amplier gain. This, however, is not an issue when comparing the rates in dierent voxels of the same grid. A count normalization was performed for each grid to the mean value. A deviation from a uniform response was found from one side of the frame to the other in grids where the blade orientation was not alternated (grids 59 to 72). This variation was found to be ∆59−72 uni =± 1%, as is shown in Figure 7.14, in good agreement with the expected value based on the coating gradient and the eciency simulation described above. Using the simple idea of alternating blade direction (frames 30 to 58) a variation of ∆30−58 uni =± 0.2% along the frames was achieved. The eect of the gradient in the boron carbide coating was reduced by a factor of 5. It should be noted that the eect of the electronics is excluded from this analysis. 124 Chapter 7. Detector Characterisation Normalised Counting 0.99 1 1.01 Pixel 1 2 3 4 Frames 30-58 Frames 59-72 Figure 7.14 : To suppress the gain dierences between grids, a normalization was performed. Variation in the eciency along each grid was measured. Variation clearly improved when the blade direction was alternated: ∆30−58 uni =±0.2 %. 7.5 Gamma sensitivity The γ -ray sensitivity of the Multi-Grid detector was measured using two different γ -ray test sources. They were placed 1.2 cm away from the square single voxel surface (2 cm × 2 cm). The solid angle coverage, Ω , was ∼ 13% of the unit-sphere surface. For each specic γ -ray source, the ux, φ , at the surface of the voxel is given by: φ=A·I·Ω (7.3) where A is the activity [ Bq ] of the source and I the intensity of the specic γ -ray under study. As an example, for the 81 keV γ -ray present in the 133 Ba source with an activity of 1.8 · 10 5 Bq and an intensity of 32.9% [50], its ux at the voxel surface is given by: φ = 1.8 · 10 5 Bq · 0.329 · 0.13 = 7.7 · 10 3 Bq. Two γ -ray test sources ( 133 Ba and 60 Co) covering a wide energy range (from tens of keV up to the MeV regime) were used to evaluate the γ -ray sensitivity. For a Conclusions The needs of neutron detectors in NSS were dened. Dierent requirements on detection eciency, uniformity, γ -sensitivity, long-term stability, etc. were studied. The present 3 He shortage prevents the use of this technology for large detector arrays. Due to their specications, ToF instruments are the most aected by the present situation. Dierent alternatives pointed out by the scientic community were analysed. Boron-lined converters coupled with gas proportional counters were chosen for this study. Dierent types of interactions occurring to the various particles involved in neutron detection (neutrons, electrons, ions and photons) were studied. Basic characteristics of gaseous detectors were reviewed. A MC simulation was performed in order to optimise several important parameters needed for the design of the proposed detector. One of this parameters was the individual layer thickness. The optimal value was found to be ≈ 1.0 µ m for a 30-layer conguration and λ = 2.5 Å. Other eects such as the wavelength dependence of the detection eciency or the wall-eect for several stopping gas pressures were simulated. The Multi-Grid concept was introduced. The dierent prototypes used in this work were study and their specications reviewed. Front-end electronics were analysed and possible optimisations pointed out. A complete study of the boron converters was done. Two dierent complementary approaches were used: thin lm characterisation and neutrography. Information on the 10 B-concentration, mass density, layer thickness, concentration of contaminants, etc. was obtained. Results helped to understand the performance of the Multi-Grid detector. A characterisation study of the Multi-Grid prototype, which included an accurate ux estimation measurement based on a 3 He-reference detector, was performed. The detection eciency was found to be n = 53.09 ± 0.54% (stat.) for incoming monochromatic neutrons of 2.5 Å. This value is in agreement with the MC simulation, which predicted an eciency of 53.47%. 132 Chapter 7. Detector Characterisation Thickness variations in the boron coating of up to 13% along the aluminium blades were measured with thin-lm characterisation techniques. This result was further conrmed by neutrography, The thickness gradient produced an eciency variation of ∆uni =± 1% along each grid. By correctly orienting these blades, an improvement was achieved. These results prove that potential constraints regarding the coating uniformity can be overcome. γ -sensitivity tests were performed for a broad range of energies (from tens of keV to a few MeV). The result obtained at the same gain value as for neutron detection eciency measurements was γ≤ 5 · 10 −5 . Long-term stability has been pointed out as one of the most important requirements for the detector of a ToF instrument. Two dierent types of gas species were tested: CF 4 and Ar-CO 2 . The Multi-Grid detector was under a moderated neutron ux over a long period of time. Small variations in the gain (measured by the shift of the PHS towards lower energies) were observed. For the moderated neutron rate used, CF 4 shows a better stability than Ar-CO 2 . If the neutron ux is further increased the loss in the detector gain becomes critical. In this sense, a clear dierence in the performance of the two gas species is found. For a similar neutron ux, CF 4 shows a decrease in gain of ≈ 80% whilst in the case of Ar-CO 2 it only drops by 10%. The relatively high eciency found, together with the high uniformity over large areas, demonstrate the potential of this technique as alternative to 3 Hebased PSDs in NSS. Bibliography [1] K. Zeitelhack. Search for alternative techniques to Helium-3 based detectors for neutron scattering applications. Neutron News , 23(4), 2012. [2] H. Mutka J. Ollivier and L. Didier. The new cold neutron time-of-ight spectrometer IN5. Neutron News , 21(2), 2010. [3] ILL patent application EP 1320119. [4] A. J. Dianoux and G. Lander. Neutron Data Booklet, 2003. [5] T. Brückel et al. Laboratory Course in Neutron Scattering . Forschungszentrum Jülich GmbH, 2007. [6] Conceptual design report. Technical report, ESS, 2012. [7] D. A. Shea and D. Morgan. The Helium-3 shortage: Supply, demand, and options for congress. Technical Report R41419, Congressional Research Service, 2010. [8] GAO. Neutron detectors: an alternative to using Helium-3. Technical Report 11-753, US Government Accountability Oce, 2011. [9] A. Cho. Helium-3 shortage could put freeze on low-temperature research. Science , 326(5954), 2009. [10] RT. Kouzes. The 3 He supply problem. Technical Report 18388, PNNL, 2009. [11] P. Convert and J. B. Forsyth. Position Sensitive Detection of Thermal Neutrons . Academic Press., 1983. [12] T. Wilpert. Boron Triouride Detectors. Neutron News , 23(4), 2012. [13] N. G. Rhodes. Scintillation Detectors. Neutron News , 23(4), 2012. 134 BIBLIOGRAPHY [14] T. Nakamura et al. A large-area two-dimensional scintillator detector with a wavelength-shifting bre readout for a time-of-ight single-crystal neutron diractometer. Nucl. Instrum. and Methods A , 686(0), 2012. [15] J. Pancin et al. Piccolo Micromegas: First in-core measurements in a nuclear reactor. Nucl. Instrum. Meth. A , 592(12), 2008. [16] J.C. Buet et al. Advances in detectors for single crystal neutron diraction. Nucl. Instrum. and Meth. A , 554(13), 2005. [17] J.L. Lacy et al. Novel neutron detector for high rate imaging applications. In Nuclear Science Symposium Conference Record, 2002. IEEE , volume 1, 2002. [18] A. Athanasiades et al. Straw detector for high rate, high resolution neutron imaging. In Nuclear Science Symposium Conference Record, 2005. IEEE , volume 2, 2005. [19] ANSI - American National Standard Institute. Evaluation and Performance of Radiation Detection Portal Monitors for Use in Homeland Security . Technical report, ANSI, 2004. [20] CRISP European Project. http://www.crisp-fp7.eu/. [21] S. Masaoka et al. Optimization of a micro-strip gas chamber as a twodimensional neutron detector using gadolinium converter. Nucl. Instrum. Meth. A , 513(3), 2003. [22] M. Calviani et al. A fast ionization chamber for ssion cross-section measurements at n-TOF. Nucl. Instrum. Meth. A , 594(2), 2008. [23] D. S. McGregor et al. Micro-pocket ssion detectors (mpfd) for in-core neutron ux monitoring. Nucl. Instrum. Meth. A , 554(13), 2005. [24] M Tardocchi et al. CadmiumZincTelluride photon detector for epithermal neutron spectroscopy - pulse height response characterisation. Nucl. Instrum. Meth. A , 526(3), 2004. [25] M. Herman. NNDC - SIGMA: Evaluated Nuclear Data File (ENDF). [26] G. F. Knoll. Radiation Detection and Measurement . John Wiley and Sons, Inc., 1999. [27] J. Ziegler. SRIM - Stopping Range of Ions in Matter. [28] M.J. Berger et al. NIST - XCOM: Photon Cross Sections Database. BIBLIOGRAPHY 135 [29] M.J. Berger et al. NIST - ESTAR: Stopping Powers and Range for Electrons. [30] F. Sauli. Principles of Operation of Multi-wire Proportional and Drift Chambers . 1977. [31] R. Santonico and R. Cardarelli. Development of resistive plate counters. Nucl. Instrum. Meth. A , 187(23), 1981. [32] A. Bolozdynya et al. Detection of thermal neutrons in cylindrical ionization chamber lled with high-pressure Xe+3He gas mixture. Nucl. Instrum. Meth. A , 522(3), 2004. [33] A. Morozov et al. Eect of the electric eld on the primary scintillation from CF4. Nucl. Instrum. Meth. A , 628(1), 2011. [34] R. Veenhof. Gareld - Simulation of Gaseous Detectors. [35] W. Shockley. Currents to conductors induced by a moving point charge. J. Appl. Phys. , 9(10), 1938. [36] S. Ramo. Currents induced by electron motion. Proceedings of the IRE , 27(9), 1939. [37] S. Chabod et al. Modelling of ssion chambers in current modeanalytical approach. Nucl. Instrum. Meth. A , 566(2), 2006. [38] J.K. Shultis and D.S. McGregor. Design and performance considerations for perforated semiconductor thermal-neutron detectors. Nucl. Instrum. Meth. A , 606(3), 2009. [39] K. A. Nelson et al. Investigation of a lithium foil multi-wire proportional counter for potential 3 He replacement. Nucl. Instrum. Meth. A , 669(0), 2012. [40] ILL patent application FR 10/51502. [41] Y. Blanc. Le spectrometre a temps de vol IN6, characteristiques, techniques at performances. Technical report, ILL, 1986. [42] O. Knotek, E. Lugscheider, and C.W. Siry. Tribological properties of B-C thin lms deposited by magnetron-sputter-ion plating method. Surface and Coatings Technology , 91(3), 1997. [43] Ali O. Sezer and J.I. Brand. Chemical vapor deposition of boron carbide. Materials Science and Engineering: B , 79(3), 2001. 136 BIBLIOGRAPHY [44] E. Pascual et al. Boron carbide thin lms deposited by tuned-substrate RF magnetron sputtering. Diamond and Related Materials , 8(25), 1999. [45] C. Höglund. Growth and Phase Stability Studies of Epitaxial Sc-Al-N and Ti-Al-N Thin Films . PhD thesis, Thin Film Physics Division - IFM - Linköping University, 2010. [46] M.U Guruz, V.P Dravid, and Y.W Chung. Synthesis and characterization of single and multilayer boron nitride and boron carbide thin lms grown by magnetron sputtering of boron carbide. Thin Solid Films , 414(1), 2002. [47] A. Lousa et al. Eect of ion bombardment on the properties of B4C thin lms deposited by RF sputtering. Thin Solid Films , 355356(0), 1999. [48] M.J. Zhou et al. Microstructure and mechanical properties of B4C lms deposited by ion beam sputtering. Thin Solid Films , 516(24), 2007. [49] C. Höglund et al. B4C thin lms for neutron detection. J. Appl. Phys. , 111(104908), 2012. [50] A. Sonzogni et al. NNDC - BNL: National Nuclear Data Center.