Study of fast radiation-detectors based on fast halide scintillator crystals and their application to the CERN n_TOF experiment

UNIVERSIDAD DE SANTIAGO DE COMPOSTELA FACULTAD DE FISICA Departamento de Física de Partículas Study of fast radiation-detectors based on fast halide scintillator crystals and their application to the CERN n_TOF experiment María de la Soledad Robles Manzano Septiembre 2016 UNIVERSIDAD DE SANTIAGO DE COMPOSTELA FACULTAD DE FISICA Departamento de Física de Partículas Study of fast radiation-detectors based on fast halide scintillator crystals and their application to the CERN n_TOF experiment María de la Soledad Robles Manzano Septiembre 2016 UNIVERSIDAD DE SANTIAGO DE COMPOSTELA Ignacio Durán Escribano, Catedrático de Física Atómica, Molecular y Nuclear de la Universidad de Santiago de Compostela, CERTIFICA: que la memoria titulada Study of fast radiation-detectors based on fast halide scintillator crystals and their application to the CERN n_TOF experiment, ha sido realizada por María de la Soledad Robles Manzano en el Departamento de Física de Partículas bajo su dirección y constituye el trabajo de tesis que presenta para optar al grado de Doctora en Física. Santiago de Compostela, a 28 de septiembre de 2016. Fdo. Ignacio Durán Escribano Fdo. María de la Soledad Robles Manzano Acknowledgements The work of this PhD has involved the effort of many people and different laboratories: both the Experimental Group of Nuclei and Particles, GENP, at the Universidad de Santiago de Compostela and the n_TOF collaboration at CERN. Firstly, I would like to thank my supervisor, Prof. Ignacio Durán for invited me to join the GENP group, specially when I was wondering why I used 152Eu instead of 137Cs and 60Co to calibrate a detector in the students lab (if you continue reading this work, you will notice that using 152Eu is still a good idea and indeed you can see nice europium spectra). Thanks Nacho for giving me this good opportunity to work in this field. I can not forget the valuable help at the n_TOF collaboration of Eric Berthoumieux, Enrico Chiaveri, Nicola Colonna, Carlos Guerrero, Frank Gunsing, Andrea Tsinganis and Christina Weiss who made possible the setups and measurements carried out at the n_TOF facility. A special mention deserved to Emilio Mendoza and Daniel Cano for their help converting the files acquired at CERN and Jorge Lerendegui and Miguel Cortés for their help providing the simulation for the n_TOF spallation target. I am very grateful to the PPACs experts: Laurent Tassan-Got, Laurent Audouin, Lou Sai Leong from IPNOrsay, Carlos Paradela from ECJRC-IRRM, Diego Tarrío from Uppsala University and Ignacio Durán and Esther Leal from USC, who provided me the PPACs ROOT files for the coincidence analysis. I specially thank Héctor Álvarez, Pablo Cabanelas and Paloma Díaz for their support with the simulations and David González for his wisdom and help when wrapping and encapsulating the crystals used in this work. I can’t forget the long and short discussions with Manuel Caamaño1 about Physics and Metaphysics, specially about the fission process. 1Give me five! VII I thank my collegues: Paloma Díaz, Noelia Montes (who guided my first steps in electronics), Martín Gascón (thanks for the scripts for the MCA and YOKO!), Yassid Ayyad (e Iris Ramos), Jossitt Vargas, Beatriz Fernández, Juan Ramón Pereira, Juan José Llerena (thanks for the spectra, all the lab stuff and the Catalina alias), Esther Leal, Diego Ramos, José Luis Rodríguez, David González, Carlos Paradela, Juan Alcántara, Diego Tarrío, Raquel Sorribas, Saúl Beceiro, Dolores Cortina, Marcos Seco (thanks for your valuable help with all my computers!), Vanesa Nimo and all the members of GENP for the stimulating discussions, for working together and for all the fun we have had in the last years. Estos años no serían tan llevaderos sin los momentos pasados fuera del laboratorio y el despacho, gracias por todo a mis amigos de Santiago: Ana F., Ana H., Bárbara, Iria, Jose C., Jose (Oscar Luis), Marijose, Paloma, Pablo, Sofía y Teresa. A mis señoras de Almería: Carmen, Cristina, María del Mar, Marijose (¡gracias por ese respiro en Annecy!), Mariló, María Rosa (y el Sr. Nacho G.) y Soraya. Gracias a Ángeles y Ramón, mis padres, que siempre me apoyan; sin ellos esto no sería posible. A mi hermano Gabriel y a mi familia, en especial a mis tíos José Luis, Maricarmen, Soraya, Gabriel y Carmen por su afecto y cariño. Y a ti, Gosha, por tu apoyo en todo, tu paciencia, los dibujos, los grandes momentos y los pequeños logros que compartimos. "There is something breathtaking about the basic laws of crystals. They are in no sense a discovery of the human mind; they just are - they exist quite independently of us-. The most that man can do is become aware, in a moment of clarity, that they are there, and take cognizance of them." -M.C. Escher 1.8. Amplitude of simulated output signa of the attenuator based on Zener diode depending on the pulse height input signal for CsI(Na)-PMT scintillator detector . . . . . . . . . . . . 19 1.9. Filter kernel when a 7th order filter is passed over the digital signal once, twice and four times. Note trends to a Gaussian kernel [2]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 1.10. Signal smoothed by a 5th order triangular filter. Raw signal is shown is the first canvas. 1 pass filter in second canvas. 2 pass filtered in third canvas. Note the smooth of the shape as the number of passes increases. . . . . . . . . . . . . . . 24 2.1. Mass stopping power [3]. . . . . . . . . . . . . . . . . . . . 28 2.2. Fractional energy loss per radiation length in lead as a function of the electron or positron energy [3]. . . . . . . . . . . 30 2.3. Photon total cross sections as a function of incident photon energy in carbon and lead [3]: σp.e. =Atomic photoelectric effect σRayleigh =Rayleigh (coherent) scattering-atom neither ionised or excited σCompton =Incoherent scattering (Compton scattering off an electron) κnuc =Pair production, nuclear field κe=Pair production, electron field σg.d.r. =Photonuclear interactions, most notably the Giant Dipole Resonance. In these interactions, the target nucleus is broken up. Original figures through the courtesy of John H. Hubbell (NIST). . . . . . . . . . . . . . . . . . . . . . . 32 2.4. Schematic representation of the electron structure in a scintillation material with a dopant, that play a role as luminescence centre. The scintillation process is also shown. Other processes involved are neglected for the sake of simplicity. . 34 2.5. Scintillator crystals used in this work: (a) CeBr3crystal encapsulated by Scionix. (b) CsI(Na) crystal. (c) CeF3 crystal L = 50 mm. Note the non-polished surface. (d) CeF3crystal L = 40 mm wrapped by two foils of ESR from 3M reflector material. . . . . . . . . . . . . . . . . . . . . . 39 2.6. Cross section of metal channel dynodes[4] (a). PMT mod. R7600U-200 from Hamamatsu (b). . . . . . . . . . . . . . . 41 2.7. Q.E. of UBA, SBA and BA photocathodes for CeBr3(a). Spectral response of PIN diode model S32004-08. Courtesy of Hamamatsu Photonics K.K. (b). . . . . . . . . . . . . . . 41 2.8. Structure of a PIN diode, (a) [5]. PIN Diode S3204-08 from Hamamatsu (b). Courtesy of Hamamatsu Photonics K. K. [6] 43 2.9. Photon Yield of CeBr3, CsI(Na) and CeF3[7],[8], [9] . . . . 43 2.10. Schematic set-up for simulations. The simulated source is sketched by the circle placed a t the origin of coordinates. The parallelepiped crystal is shown as well in the figure. The dash-line represents the solid angle from the source to the end of the crystal. . . . . . . . . . . . . . . . . . . . . . 46 2.11. Detection efficiency of CeBr3(left) and CsI (right). The three simulation configurations are shown in both cases. . . 48 2.12. Mode value (left) and mean value (right) for CeBr3. The three simulation configurations are shown. . . . . . . . . . . 50 2.13. Mode value (left) and mean value (right) for CsI. The three simulation configurations are shown. . . . . . . . . . . . . . 51 2.14. CeBr3bare crystal. Energy deposited spectra at different initial energy for γ-rays. The counting efficiencies are shown for each energy . . . . . . . . . . . . . . . . . . . . . . . . . 52 2.15. CeBr3bare crystal. Energy deposited spectra at different initial energy for electrons. . . . . . . . . . . . . . . . . . . . 53 3.1. The 152Eu spectrum (in this case with the CeBr3-PMT detector). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 3.2. Calibration energy (left) and energy resolution (right) for CeBr3-PMT . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 3.3. 137Cs+60Co spectra using CeBr3-PMT detector (a). Electronic contribution resolution (b). . . . . . . . . . . . . . . . 58 3.4. 88Y+AmBe spectrum acquired using CeBr3-PMT detector. 59 3.5. Background radiation spectrum measured using CeBr3-PMT detector. No intrinsic activity is observed. . . . . . . . . . 60 3.6. 90Sr spectrum. Note the two contribution of the desintegration branches at 546 keV and 2284 keV, after passing the Al wall. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 3.7. Calibration energy (left) and energy resolution (right) for CsI(Na)-PMT . . . . . . . . . . . . . . . . . . . . . . . . . . 62 3.8. CsI(Na)-PMT 137Cs+60Co (a) and 88Y+AmBe spectra (b). 63 3.9. Calibration energy (left) and energy resolution (right) for CsI(Na)-PIN. . . . . . . . . . . . . . . . . . . . . . . . . . . 65 3.10. 137Cs+60Co spectra acquired using PDCsINa detector. . . . 66 3.11. 238Pu spectrum using CeF3_L40-PMT for different set-ups. The 137Cs spectrum is added as deposited energy reference. 67 3.12. Centroid temperature dependence CsI(Na)-PMT detector. . 68 3.13. R temperature dependence CsI(Na)-PMT detector. . . . . . 69 3.14. Temperature variation over the time for CsI(Na)-PIN detector. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 3.15. Resolution variation over the temperature for CsI(Na)-PIN detector. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 3.16. Temperature variation over the time for CeBr3-PMT detector. 71 3.17. Resolution variation over the temperature for CeBr3-PMT detector. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 3.18. CeBr3and CsI(Na) decay times. . . . . . . . . . . . . . . . 73 3.19. CeBr3-PMT profiles for γ-rays, βparticles and cosmic-rays. 74 3.20. CsI(Na)-PMT profiles for γ-rays, βparticles and cosmic-rays. 75 3.21. CeBr3-PMT timing resolution distribution. . . . . . . . . . 76 3.22. Original pulse from CsI(Na)-PMT (black) and smoothed pulse (orange). . . . . . . . . . . . . . . . . . . . . . . . . . 77 3.23. Rising time for a CsI(Na)-PMT pulse. Upper Fig. Original signal. Middle Fig. Signal after one pass 5th order triangular smooth. Lower Fig. Signal afer two pass 5th order triangular smooth. . . . . . . . . . . . . . . . . . . . . . . . 78 3.24. CsI(Na)-PMT timing resolution distribution. . . . . . . . . 78 3.25. 137Cs and 60Co spectrum obtained after a smoothing of CsI(Na)-PMT signals. The energy resolution for 662 keV R(%) = 6.7±0.1 . . . . . . . . . . . . . . . . . . . . . . . . 79 3.26. Raw signal from CeBr3-PMT (above). Filtered signal using a trapezoidal filter (below). . . . . . . . . . . . . . . . . . . 80 3.27. FIR filter 137Cs spectra obtained from the filtered signal of CeBr3-PMT using a trapezoidal filter. R(%) = 8.2 ±0.1 . . 80 3.28. CeF3_L40-PMT profiles of a sets, one for αparticles, one for γ-rays, of digitised signals from CeF3_L40-PMT detectors. Fig. (a) shows the profiles of both sets normalised to the area of each profile shape. Fig. (b) shows shows the profile of both sets normalised to the amplitude of each profile shape. . . . . . . . . . . . . . . . . . . . . . . . . . . 82 3.29. (a) Slope as a function of the integral area event-by event. (b) Mean slope value as a function of integrated area. Note the splitting between αand γ-rays samples. . . . . . . . . . 83 4.1. n_TOF facility . . . . . . . . . . . . . . . . . . . . . . . . . 89 4.2. Lead spallation target. . . . . . . . . . . . . . . . . . . . . 89 4.3. Evaluated neutron flux for different spallation targets in the EAR1 [10]. . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 4.4. Beam line EAR1. . . . . . . . . . . . . . . . . . . . . . . . 90 4.5. Beam line EAR2. . . . . . . . . . . . . . . . . . . . . . . . 92 4.6. PPAC schematic view. An anode in between of two stripped cathodes and their read-out electronics are represented. Note the induce charged in the three planes. [11] . . . . . . . . . 96 4.7. Schematic working of a PPACs detectors . . . . . . . . . . . 96 4.8. Above: PPAC signal from an anode. Below: Filtered PPAC signal used for the analysis [12]. . . . . . . . . . . . . . . . . 98 4.9. Neutron energy vs time of flight. Note the difference at 1 µs between the relavistic (blue line) and the classical formula (violet line) (a). Neutron energy as a function of the time of flight in the region of interest to detect prompt γ-rays from neutron induced reactions. . . . . . . . . . . . . . . . . . . . 100 4.10. Neutron capture set-up at EAR1 . . . . . . . . . . . . . . . 101 4.11. Trapezoidal-filtered signals for CeBr3-PMT and CsI(Na)- PMT. In (a), the blue line corresponds to a couple of the events from CeBr3-PMT registered by the n_TOF DAQ. The red line is the filtered signal, showing a constant time offset. In (b), an original signal from CsI(Na)-PMT (black line) shows a baseline shift. The green line is the smoothed original signal showing a better shape. The red line corresponds to the trapezoidal-filtered signal. Note the baseline shift correction. . . . . . . . . . . . . . . . . . . . . . . . . . 102 4.12. 197Au neutron capture cross section for 197Au. Red line: ENDF/B-VII.1 evaluation. Black line: measured scaled for comparison with the evaluation. (a) corresponds to CeBr3- PMT measurements, (b) corresponds to CsI(Na)-PMT measurements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 4.13. natAg neutron capture cross section for 197Au. Red line: CENDL-3.1 evaluation. Black line: measured scaled for comparison with the evaluation. (a) corresponds to CeBr3- PMT measurements, (b) corresponds to CsI(Na)-PMT measurements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 4.14. Measured counting rate distributions corresponding to the 197Au and the dedicated background measurements. Fig. taken from Ref. [13]. . . . . . . . . . . . . . . . . . . . . . 106 4.15. PPAC targets . . . . . . . . . . . . . . . . . . . . . . . . . . 108 4.16. Set up PPACs, scintillators . . . . . . . . . . . . . . . . . . 109 4.17. Plateaus for CeBr3-PMT (a) and CsI(Na)-PMT (b). The statistics for each timing window is shows (black dots). The flatter contribution was subtracted by fitting a linear function. The result is plotted (blue dots). . . . . . . . . . . . . 112 4.18. Types of events that happen in PPACs detectors. Type 3 and 4 must be remove from the analysis to avoid the distorted measurements. [14] . . . . . . . . . . . . . . . . . . . 113 4.19. Schematic representation of the coincidence algorithm. The dashed lines represent coincident signals between the crystal and the PPAC detectors. The dotted lines show the coincidence among PPACs. Each coincident measurement is stored using a specific integer number. This value later can be used to select via a binary mask one specific coincidence condition [14]. . . . . . . . . . . . . . . . . . . . . . . . . . . 114 4.20. Distribution of the time differences between a PPAC anode and a scintillator detector for CeBr3-PMT (a) and CsI(Na)- PMT (b). . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 4.21. 234U(n,γ) contribution. . . . . . . . . . . . . . . . . . . . . . 116 4.22. γfission neutron energy . . . . . . . . . . . . . . . . . . . . 117 4.23. g-flash signals produced . . . . . . . . . . . . . . . . . . . . 118 4.24. CeBr3-PMT and CsI(Na)-PMT detectors placed at EAR1. . 121 4.25. γ-flash pulses from CeBr3-PMT during the 2012 detector test for γ-flash. . . . . . . . . . . . . . . . . . . . . . . . . . 122 4.26. Amplitude distribution of the γ-flash for the first and second pulses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 4.27. γ-flash profiles for differen deposited energy ranges (related to amplitude distribution in Fig. 4.26) (a) Single pulses profiles. (b) Two or more pulses profiles. (c) γ-ray profile. Note the width of the signal produced by the γ-flash single contribution is larger than the a gamma-ray pulse. . . . . . 124 4.28. Timing distribution of the γ-flash for the first and second pulses. Three component are distinguished: very prompt, prompt and delayed. . . . . . . . . . . . . . . . . . . . . . . 125 4.29. γ-flash pulses from CeBr3-PMT (a) and CsI(Na)-PMT (b) during the 2012 PPAC fission measurements for γ-flash. . . 126 4.30. γ-flash pulses from CsI(Na)-PMT. Note the rebound after the decay time of the saturated signals. . . . . . . . . . . . . 127 4.31. γ-flash attenuated pulses from CsI(Na)-PMT using a Zener diode based attenuator. Note that the rebound after the decay time of signals is removed. . . . . . . . . . . . . . . . 127 4.32. Setup for γ-flash in EAR2, showing the detectors used. . . . 128 4.33. Signal profiles . . . . . . . . . . . . . . . . . . . . . . . . . . 129 4.34. γ-flash signals from CVD Diamond detector acquired in EAR1 (left) and EAR2 (right). [15] . . . . . . . . . . . . . . 130 4.35. γ-flash signals from PPAC detector acquired in EAR1, above, and EAR2, below. . . . . . . . . . . . . . . . . . . . . . . . 131 4.36. GEANT4 simulations . . . . . . . . . . . . . . . . . . . . . . 132 6.1. Espectros adquiridos de 137Cs+60Co usando las tarjetas electrónicas stretcher para la adaptación de las señales del CeBr3, (a), y el CsI(Na) (b). Los problemas de no-linealidad han sido resueltos. . . . . . . . . . . . . . . . . . . . . . . . . . . 140 6.2. Típica cadena de medida para la medida de la energía depositada, llamada cadena espectroscópica. Está compuesta por cinco etapas: el sistema de detección, una etapa preamplificadora, un amplificador que adapta la forma de la señal, el amplificador espectroscópico y por último, el sistema de adquisición. La evolución y modificación de la señal se representa esquemáticamente en la línea. . . . . . . . . . . . . . 141 6.3. Intrinsic efficiency comparison for bare CeBr3and CsI crystals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 6.4. Most probable deposited energy value (mode) and mean deposited energy value (mean) for CeBr3for electrons at different initial energies. . . . . . . . . . . . . . . . . . . . . 143 6.5. Espectro 238Pu usando CeF3para diferentes configuraciones de la medida. El espectro de 137Cs se usa como referencia para la energía depositada. . . . . . . . . . . . . . . . . . . . 146 6.6. Espectro de 137Cs obtenido a partir de la señal filtrada por un filtro FIR trapezoidal del detector CeBr3. . . . . . . . . 147 6.7. Esquema de la configuración de las PPACs donde los blancos se encuentran intercalados. . . . . . . . . . . . . . . . . . . 149 6.8. Señales producidas por el γ-flash para diferentes detectores: Micromegas, C6D6 y BaF2. . . . . . . . . . . . . . . . . . . 150 A.1. Left: Hamamatsu E5996 Voltage Divider. Right: modified version of the original Hamamatsu design to make it suitable for the very fast signals produced by high L.Y. crystals. . . 155 A.2. Deviation from linear behaviour in normal and tapered voltage divider circuits. The linear response is maintained for one order of magnitude more for the tapered design. Courtesy of Hamamatsu Photonics K.K. . . . . . . . . . . . . . . 156 B.1. Frequency response of an ideal capacitor.Courtesy of Murata Electronics [16]. . . . . . . . . . . . . . . . . . . . . . . 158 B.2. Equivalent circuit representation of an actual capacitor. Courtesy of Murata Electronics [16]. . . . . . . . . . . . . . . . . 158 B.3. Frequency response of an actual capacitor. Courtesy of Murata Electronics [16]. . . . . . . . . . . . . . . . . . . . . . . 158 C.1. Stretcher circuit design for CeBr3-PMT . . . . . . . . . . . 160 C.2. Signal at different stages of the conditioning circuit simulated using PSpice for CeBr3-PMT. . . . . . . . . . . . . . . 164 C.3. Stretcher design circuit for CsI(Na)-PMT. . . . . . . . . . . 165 D.1. Attenuator desing based on Zener diode mod. DZ2W047 from Panasonic. . . . . . . . . . . . . . . . . . . . . . . . . . 167 F.1. Event visualiser for the EnsarRoot framework[17]. A scintillator crystal is 3D sketched and the traces for different particles when an incident γ-ray hits the crystal are distinguished. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172 G.1. CeBr3bare crystal. Energy deposited spectra at different initial momentum. Incident particles: γincident. . . . . . . 173 G.2. CeBr3bare crystal. Energy deposited spectra at different initial momentum. Incident particles: e−incident. . . . . . 174 G.3. CeBr3bare crystal+Al foil 1.5 mm. Energy deposited spectra at different initial momentum. Incident particles: γ incident . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 G.4. CeBr3bare crystal+Al foil 1.5 mm. Energy deposited spectra at different initial momentum. Incident particles: e− incident . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176 G.5. CeBr3bare crystal+Al foil 5 mm. Energy deposited spectra at different initial momentum. Incident particles: γincident 177 G.6. CeBr3bare crystal+Al foil 5 mm. Energy deposited spectra at different initial momentum. Incident particles: e−incident178 G.7. CsI(Na) bare crystal. Energy deposited spectra at different initial momentum. Incident particles: γincident . . . . . . 179 G.8. CsI(Na) bare crystal. Energy deposited spectra at different initial momentum. Incident particles: e−incident . . . . . 180 G.9. CsI(Na) bare crystal+Al foil 1.5 mm. Energy deposited spectra at different initial momentum. Incident particles: γ incident . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 G.10.CsI(Na) bare crystal+Al foil 1.5 mm. Energy deposited spectra at different initial momentum. Incident particles: e−incident . . . . . . . . . . . . . . . . . . . . . . . . . . . 182 G.11.CsI(Na) bare crystal+Al foil 5 mm. Energy deposited spectra at different initial momentum. Incident particles: γ incident . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183 G.12.CsI(Na) bare crystal+ Al foil 5 mm. Energy deposited spectra at different initial momentum. Incident particles: e− incident . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184 List of Tables 2.1. Main characteristics of scintillator crystals . . . . . . . . . . 37 2.2. Crystals dimensions . . . . . . . . . . . . . . . . . . . . . . 37 2.3. Photodetectors biasing. . . . . . . . . . . . . . . . . . . . . 42 3.1. Energy resolution for CeB3-PMT, CsI(Na)-PMT and CsI(Na)- PIN scintillator detectors. . . . . . . . . . . . . . . . . . . . 85 3.2. Energy resolution for CeF3_L40-PMT and CeF3_L50-PMT detectors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 4.1. Trapezoidal filter parameters for CeBr3-PMT and CsI(Na)- PMT detectors used in promt γ-rays mesurements at n_TOF.103 6.1. Main characteristics of scintillator crystals . . . . . . . . . . 138 6.2. Crystals dimensions . . . . . . . . . . . . . . . . . . . . . . 138 A.1. Voltage dividers distribution from Hamatsu E5996 socket and the modified design by USC. . . . . . . . . . . . . . . . 154 XXV 1. Hardware and Software solutions for fast scintillator detectors Channel 0 1000 2000 3000 4000 5000 6000 7000 8000 Counts 0 500 1000 1500 2000 2500 3000 3500 4000 . 3 Co. C.A.E.N. 968 Spectrospcopic Amplifier. CeBr 60 Cs+ 137 (a) Channel 1000 2000 3000 4000 5000 6000 7000 8000 Counts 0 100 200 300 400 500 3 Co. CANBERRA 2022 Spect. Amp. CeBr 60 Cs+ 137 (b) Channel 0 200 400 600 800 1000 Counts 0 100 200 300 400 500 600 3 Y spectrum. ORTEC 572. CeBr 88 Cs+ 137 (c) Figure 1.1.: The non-linear behaviour can be appreciated when the signals were not previously conditioned. 137Cs+60Co ((a) and (b)) and 137Cs+88Y (c) spectra measured with different standard spectroscopic amplifiers. 6 1.1. Hardware solutions 1.1. Hardware solutions The challenge of dealing with a very high current peak produced in the optodevices by high-performance fast scintillator crystals, requires specific hardware solutions. Concerning PMTs, it was already commented the relevance of the design of the VD chain and how it influences the signal fidelity regarding the original shape from the scintillator. Furthermore, the sooner the signal is well shaped without loosing information on the relevant observables, the lower the undesired effects should be produced in the rest of the front end electronic chain. The first temptation one has when dealing with such a fast crystals is lowering the HV in order to reduce the gain of the PMT. But reducing the gain results, in any case, in a reduction of the spectroscopic resolution and if the HV reduction is severe the resolution deteriorates even more by the fact of a bad focusing of the photo-electrons in the first stage. Some PMT manufactures suggest to use a Zener-diode in the first link of the VD chain, instead of a resistor, allowing so a certain margin of gain control without de-focusing the first stage. Other PMT manufacturers produce active chains, having transistors in the three last links of the chain, improving so the linearity for large pulses. Some users decide to pick-up the signal from the next to last dynode and reducing in consequence the load resistor at the anode, leading so to a better behaviour of the electron swarm in the last stage. Finally, there is the possibility of using an active circuit to reduce the gain in the last stages when a big scintillation is detected in the first ones. Most of these remedies achieve very limited success when the raising edge of the scintillation is lower than 1 ns and very often create a ringing effect in the rest of the pulse. An easy solution, used in this work with the CeBr3crystals, is to modify the VD by reducing the voltage in the last stages without changing it at the first ones. To do this, the penultimate and the ante-penultimate resistors were reduced by a factor of two and, besides this, reducing the HV in consequence to get a gain’s reduction without affecting the focussing potential at the first dynode. In addition to this, when dealing with ultrafast crystals, a commonly useful trick is to put a capacitor in parallel 7 1. Hardware and Software solutions for fast scintillator detectors with the load resistor. This capacitor acts reducing the impedance of the anode’s load mainly for the higher frequencies. It is worth discussing on the convenience of reducing the high frequency components by a capacitor in parallel with the anode’s load. As a first advice one must refrain from using components in the VD (neither resistors, capacitors, diodes nor transistors) not being specified for frequencies well beyond the GHz range (normally, the SMD components used in fast electronics are compliant with, but most of the resistors and capacitors found in the shelves are not). Should the appropriate capacitors been used, the first observed effects are the slowing down of the rising time and the reduction of the peak amplitude. These effects seem to worsen the quality of the output pulse (both time and amplitude resolution should be affected) but this is not the case when both the extremely short rise time and the extremely high amplitude are creating a bigger trouble. In fact, the finally got time resolution depends on both the rise time and the time resolution of the Front End Electronics (FEE), in such a way that sending extremely fast signals to the FEE, paradoxically, does not result in a better time resolution and results in a poorer energy resolution. This is partly due to the fact of the higher frequency components of the signal not being well managed neither by the cables nor by the FEE. Concerning the amplitude of the pulses, the use of a well suited capacitor does not entail a loss of information on the energy deposited in the crystal, because the PMT’s output is a current signal and all the charge integrated in the capacitor is always given back, provided we are using low loss HF components. Should the RC time constant become well adapted to the signals we are dealing with, both time and amplitude resolutions rest undamaged. Problems produced by fast signals in cables and Front-End Electronic, FEE. The main effect produced in cables is a worsening of the Signal-to-Noise Ratio (SNR). Obviously, the longer the cable is the worst the SNR will be. Firstly because of the attenuation of the Very High Frequency (VHF) components, and secondly because cables are a source of noise, even if coaxial ones were used. The connectors type and the electrical specifications of the cables take part in the propagation of the signal in an appropriate way, i.e., neither 8 1.1. Hardware solutions producing loses nor modifying its shape. Most of the cables used in the nuclear experiments have loses. The larger the frequency is, the more the loses causes by propagation are. That means that signals whose frequency is less that decades of nanoseconds, the loses by propagating along the cable will be larger. The attenuation is produced by the non-linear behaviour of the cable impedance, being function of the signal frequency spectrum. The wire resistance is affected by a spurious inductance and the capacitance by the so-called dielectric that insulates the wire and the grounding. Actually, the insulator is made of a material having a non-linear behaviour, neither in frequency nor in amplitude, in such a way that the nominal capacitance appears as having a non-linear ohmic resistance to ground. This is degrading so both the amplitude and the shape of the original signal. The specifications of cables and connectors must so match the shaping of the fast-detector output, and SMA standard connectors and cables must be used instead of LEMO or BNC standard ones, when required. Once again, the advantage of dealing with signals not having frequency components above few GHz becomes obvious. A second effect due to coaxial cables is the fact of they are linking the ground of both the sending signal system and the receiving one. It is worth noting that all the current flowing in one direction through the signal wire must return via the cable ground mesh. This is specially relevant when dealing with a huge number of fast signal channels, because they produce peaks of current that disturb the ground system. Of course, the higher the frequency of the signal is, the more severe are the problems created in the ground systems. A commonly proven solution is the use of twistedpair cables were both wires are floating and so there is no ground links. Moreover, in the twisted-pair configuration, both wires are picking-up the same noise, that cancels when reading in a differential mode. The use of HF twisted-pair cables is highly recommended, as far as possible. 1.1.1. Voltage Supply Network PMTs having a high Q.E. and a high gain, when coupled to high L.Y. CeBr3crystals produce a high current signal peaks because of its short 9 1. Hardware and Software solutions for fast scintillator detectors Figure 1.2.: Response curve of a PMT as a function of the Light Yield. Ideally the response is linear, shown with a dashed line. The solid line shows the non-linear relation for high Light Yield. (Figure courtesy of Hamamatsu Photonics K. K. [1]). rising time and very fast decay component (see Table 2.1) that create a huge number of photoelectrons in the photocathode in a very short period of time. As a consequence it was commented above, the dynode stages saturated leading to a degradation in the output linearity. To solve this problem a new VD was designed. To assure a low SNR factor, this development was done based on D Type Socket Assembly Normal Divider mod. E5996 from Hamamatsu [23] (see Fig. A.1 (left) in Appendix A). 1.1.2. Bias voltage curve Hamamatsu Photonics provides a exponential correlation of gain versus the voltage supplied (see Fig.1.3) to the PMT R7600U-200 for the standard VD circuit (see ref [23]). The newly designed VD circuit, requires a new calibration and two measurement methods have been carried out for this purpose: The first method consists on the acquisition of the detector output signal 10 1.1. Hardware solutions (V) bias V 400 500 600 700 800 900 gain 500 1000 1500 2000 2500 3 10× / ndf 2 χ 1.883e+10 / 105 Constant 0.03822± 3.939 Slope 4.302e-05± 0.01158 Figure 1.3.: Gain vs Voltage Supply of the PMT R7600U-200+E5996 socket. Courtesy of Hamamatsu Photonics K.K. directly connected to a YOKOGAWA oscilloscope [24]. A large amount of pulses were acquired for every value of a set of supplied voltages so long as the CeBr3-PMT was irradiated by a 137Cs radioactive source. Clearly, the amplitude of the signal is proportional to the charge produced in the PMT, and therefore, to the gain. Then, the waveform of every pulse was analysed using a digital algorithm, (which will be explained in detailed in Chapter 2) to find out the relation between the maximum amplitude of every acquired pulse and the supplied voltage supplied to the PMT. The observable maximum amplitude, related directly to the PMT gain, and the voltage supplied correlation is proportional to the gain vs. voltaged supplied correlation provided by the manufacturer, following an exponential dependence. See Fig. 1.4 (a). The second method is based on the correlation between the centroid of the 137Cs radioactive source spectra acquired using a spectroscopic measurement chain (that will be explained in detail in following sections in Chapter 1) consisting on several active stages: an electronic preamplifier, an spectroscopic amplifier mod. 2022 manufactured by CANBERRA and 11 1. Hardware and Software solutions for fast scintillator detectors (V) bias V 580 600 620 640 660 680 700 720 Mean (V) 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 / ndf 2 χ 14.45 / 11 Constant 0.05856± -9.628 Slope 8.86e-05± 0.01183 / ndf 2 χ 14.45 / 11 Constant 0.05856± -9.628 Slope 8.86e-05± 0.01183 (a) (V) bias V 600 620 640 660 680 700 720 Channel (ch) 500 1000 1500 2000 2500 / ndf 2 χ 4304 / 7 Constant 0± -0.9522 Slope 8.661e-06± 0.01225 (b) Figure 1.4.: Maximum pulse amplitude vs supply voltage of the PMT R7600U-200+USC-VD, measured using PSA of signals recorded with a YOKOGAWA oscilloscope (a). Centroid for 662 keV γrays froam a 137Cs vs supplied voltage measured with a spectroscopic measurement chain: preamplifier, shaper, spectroscopic amplifier and multichannel analyser (b). a MCA from AmpTek. This analysis of the centroid of the photopeak of 662 keV γ-rays from 137Cs. confirms the exponential dependence as well. See Fig. 1.4 (b). 1.1.3. Spectroscopic chain measurements In nuclear physics experiments, besides simply detect the production of a certain events (counters), one could need to measure some specific observable magnitudes such as, for instance, the time (timers), the deposited energy (spectroscopic analysers), the signal shape (Pulse Shape Analysers, PSA), or any combination of them. The electronic pulses must so be adequately handled in order to preserve its characteristics related to the observables of interest. When dealing with spectroscopic analysers, a typical set-up consists on mainly five different stages: the detection system, the preamplification stage, the shaping amplifier, the spectroscopic amplifier (SA) and the acquisition system, as sketched in Fig. 1.5. 12 1.1. Hardware solutions The detector produces either a voltage or a current signal (see Refs. [19, 25]) or very often something intermediate, and so the preamplifier must be specifically chosen to match the detector output. A typical voltage signal should be coupled to an extremely high-impedance input amplifier whereas a typical current signal will require a low-impedance input to drain the charge produced by the detector. Commonly, the output impedance of the detector is a complex impedance and so the preamplification stage can be seen as an active filter, adapting the input/output impedances (transimpedance). It must be placed as close as possible to the detection system to avoid unwanted effects due to the stray capacitances of cables. Shaping amplifiers are used to amplify the signal modifying its shape to make it appropriate to the rest of FEE. Tipically, the SA’s, on top of its amplitude gain, manage parameters such as the polarity, the timing, the shape, and also perform a baseline restorer in order to condition it for the down stream spectroscopic ADQ system. The spectroscopic systems usually consists on a MultiChannel Analyser (MCA) or an Analog to Digital Converter (ADC), as it will be shown in Chapter 3 and Chapter 4. Figure 1.5.: Typical measurement chain for measuring the deposited energy, called spectroscopic chain. It consists on five stages: the detection system, the preamplifier stage, the shapery amplifier, the spectroscopic amplifier and finally, the acquisition system. The evolution and modification of the signal is schematically represented in the line. 13 1. Hardware and Software solutions for fast scintillator detectors 1.1.3.1. Non-linearity effects To overcome the problem of lack of linearity when the time characteristics of the analog signal from the detectors were not matching the specifications of the SA and the MCA, a hardware solution is proposed. Since the non-linearity effects observed are due to a problem of signal conditioning, the characteristics of the output signal from the optodevices coupled to the scintillator detectors must be cope with the specifications of the spectroscopic amplifier (SA); otherwise, when designing the electronic chain, an specific shaping stage must be implemented. The spectroscopic amplifier chosen for that purpose is the CANBERRA Mod. 2022 whose main requirements for the input signals are[20]: 1. Signal polarity: positive or negative pulses. 2. Rise time less than shaping time chosen: <0.5 µs, 1 µs, 2 µs, 4 µs, 8 µs or 12 µs. 3. Decay time larger than 40 µs for 0.5 µs, 1 µs, 2 µs, 4 µs, 8 µs or 100 µs for 12 µs. 4. Amplitude ±10 V divided by the selected gain for linear response. Regarding the output signals from both CeBr3-PMT and CsI(Na)-PMT scintillator detectors in Figs. 3.18 and ?? respectively, it is noticeable that both decay times -around 30 ns and 600 ns respectivelyare below the required by the SA CANBERRA Mod. 2022 specified in the requirement 2. Stretcher Printed-Circuit-Board design To be consistent with the discussion above, the following considerations were taken into account when designing the shaping circuit, that in following is called stretcher: 1. The conditioned signal must maintain proportional to the observables to the initial signal. 2. The decay time of the conditioned signal must be at least 500 ns. 3. The used components must be appropriate for the frequencies of the initial signal and the subsequent ones (see Appendix B). 14 1.1. Hardware solutions The designs of the both strecher PCBs, one for CeBr3-PMT another for CsI(Na)-PMT, were simulated using the software package OrCAD 10.5 [26] including, among other programs, PSpice, Capture CIS and Layout Plus. A detailed explanation of the designs can be read in Appendix C. Fig. 1.6 shows the strecher PCBs for CeBr3-PMT detector. The non-linear spectra, badly obtained without its specific signal conditioning stage, avoid an appropriate characterisation of the response of the scintillator detectors in terms of resolution and linearity. Concerning the energy resolution, it can be calculated by the Eq. 1.1 R(E) = FWHM E×100 ≡2.35σ E×100 (1.1) where E is the energy deposited in the detector, the Full Width High Maximum (FWHM) must be given in the same units as E and is related to the standard deviation for peaks whose shape is Gaussian, as it is most often the case of the spectra peaks from radioactive sources. Consequently, good linearity is needed to provided an accurate value of the energy resolution as it was explained above. The different stretcher PCBs were done to solve these problems, conditioning the output of the scintillator detectors by taking into account the requirements of the SA. Both stretcher PCBs, for CeBr3-PMT and CsI(Na)-PMT, provide linearity after the signal passing the SA, solving the problem discussed previously. In Fig. 1.7 a 137Cs and 60Co spectra for CeBr3-PMT and CsI(Na)-PMT respectively are shown, without having the non-linear effects present in Fig. 1.1 are solved. 15 1. Hardware and Software solutions for fast scintillator detectors y[n] = (f∗g) = ∞ X m=−∞ f[n−m]·g[m](1.3) where fis the convolution function, commonly called kernel, and g is the input. Concerning FIR filters, function fis replaced by a set of N coefficients, hi, where N is a natural number that defines the order of the filter. These coefficients dictate the weight of every sample included in the order of the filter. Thus, the convolution of the FIR filters is as follows: y[n] = N X i=0 hi·g[n−i](1.4) The choice of the order of the filter must be made carefully, taking into account the number of samples of the entire digitised signal. One the one hand, in case the number of samples is close to the order of the filter, the output will be reduced in samples, having N-n samples, and the information contained in the signal might be lost. On the other hand, if the order is smaller than it should be, the output might not fulfil the requirements. Another important filter parameter is its width, that influences amplitude and delay of the output. Moreover, the choice of the values of the coefficient and their sign, defines different filter types and behaviours. For instance, two filters whose absolute values of the coefficients are the same, but their sing are the opposite, may implement either an integrative or derivative effect. In the following, some of such FIR filters are going to be introduced. Moving Average Filter The moving average filter is one of the most simple FIR filters. It operates by averaging a specific number of samples (bins), related to the order of the filter, of the digitised signal. y[n] = 1 L L−1 X i=0 g[n+i](1.5) Seeing Eq. 1.5, the moving average filter is a convolution of the digitised signal with a rectangular pulse having an area of one. 22 1.2. Algorithms for digital signal processing It may be implemented by two options: •The filter considers a number of samples L forward to calculate the filtered signal, reducing so by L the final number of samples. •The filter considers for the ith sample the same number of samples backwards and forward, that implies an odd number of coefficients, this is commonly called symmetric moving average filter, reducing or not the number of final samples. Triangular smooth: tends to Gaussian function. Triangular or Bartlett FIR filter are commonly applied as a smooth of signals or spectra that present high frequency noise (see Fig. 1.10). The most commonly used triangular filter are the 3rd, 5th and 7th order, which implies a set of 3, 5 and 7 coefficients respectively, they are calculated according the Eq.(1.6) where, in order to assure a unit-gain smoothing the sum of the coefficients must be equals one. n+ 1 (N−1 2+ 1)2if n ∈0,N 2 N−n (N−1 2+ 1)2if n ∈N 2, N −1(1.6) A remarkable feature of these filters is that recursive application, twice or more, results in a higher order smooth tending to a Gaussian filter kernel as it can be seen in Fig. 1.9 Trapezoidal filter This filter is called trapezoidal due to the shape of the resulting output signal when it is applied to a step function, however both the low and high constant values are separated not by an abrupt step but by a steeped slope. The filter is formed by two moving average filters separated by a time gap, G, (see Eq. 1.7) usually shorter than the rise time of the signal to be filtered. y[n] = 1 L L−1 X i=0 g[L+G+n+i]− L−1 X i=0 g[n+i]!(1.7) 23 1. Hardware and Software solutions for fast scintillator detectors Figure 1.9.: Filter kernel when a 7th order filter is passed over the digital signal once, twice and four times. Note trends to a Gaussian kernel [2]. Figure 1.10.: Signal smoothed by a 5th order triangular filter. Raw signal is shown is the first canvas. 1 pass filter in second canvas. 2 pass filtered in third canvas. Note the smooth of the shape as the number of passes increases. 24 1.3. Conclusions on the hardware and software solutions proposed Hann filter The Hann (or Hanning) filter has a sinusoidal shape, a good frequency resolution and can be used in most cases because of its good continuity properties that reduces the spectra leakage. The Hann filter may cause a distortion of the initial signal and therefore some corrections should be done to obtained an accurate result [28, 29]. The coefficiencies of a Hann filter f(t) = 1 2+1 2cos 2πn N−1,0≤n≤N(1.8) 1.3. Conclusions on the hardware and software solutions proposed The fast signals produced by fast scintillator detectors must be handle taking into account the problems related to its digital signal processing by the acquisition system used. For high L.Y. crystals, the photodetector output must be adapted to avoid saturation problems. Solutions have been proposed by means of the modification of the original VD in order to adapt the high amount of light produced by CeBr3crystal. The supply voltage to the dynode stages of the PMT were modified by simply adding a few components in parallel to the original ones. To obtain an accurate energy resolution and an appropriate signal conditioning two electronic circuit, one for CeBr3-PMT and one for CsI(Na)-PMT, have been designed and developed. To avoid saturation problems in fADC systems when signals overcome the input signal amplitude accepted by the system, an ad-hoc attenuator were designed and tested. The simulation, measurements and results obtained indicate that the solutions proposed solved the problems found when dealing with fast scintillation crystals. Good linearity response of the detector in terms of energy resolution, calibration and non-saturation were obtained after having adapted appropriately the signals to the requirements of the acquisition systems. 25 1. Hardware and Software solutions for fast scintillator detectors 26 2. Scintillator detectors Scintillator detectors are based on the property of some kind of plastics and crystals of producing and transmitting light when energetic charged particles slow down by ionising or exciting its electronic states. It is worth to recall that energetic enough photons ( X- or γ-rays) can also produce free electrons, being in this way detected inside the scintillator volume. The amount of subsequent scintillation-photons depends so on the total energy deposited by the incident radiation, being the dynamics of the secondary processes depending also on the radiation’s kind, and so, leading to different shapes of the scintillation pulse. A deep analysis of these pulse parameters will give the access to valuable information about the original physical processes. To a better understanding of the results of the measurements performed in Chapter 3 and Chapter 4 is needed to bear in mind the basics principles of the radiation interaction in matter and of the scintillation processes. For a deeper comprehension of the detectors used in this work, a realistic Monte Carlo simulation of the deposited energy by both electrons and photons is also included in this Chapter. 2.1. Interaction of radiation with matter The energy loss due to collisions with electrons by a charged particle in a given material, also called stopping power, was firstly calculated classically by Bohr [30] and the Bethe’s theory included then corrections for quantum effects [31]. This theoretical formula, depending on Z, the electron density of the stopping material and a parameter, I (called "mean excitation potential"), was successful but showed deviations from the experimental data, at energies below around 200 keV. Bloch [32] introduced corrections, leading to the so called Bethe-Bloch formula, even though that further 27 2. Scintillator detectors corrections have been introduced later (see Appendix E). 2.1.1. Energy loss by heavy charged particles Heavy charged-particles loss their energy mainly by the Coulomb interaction with the bound electrons of atoms of the material. In such interactions, the atoms can be either excited or ionised, depending on the energy of the charged particle. In the region of interest of this work, the ionisation process is dominant, as the energy of the incident particle is much larger than the bound energy of the atomic electrons. The Bethe-Bloch formula describes well for 0.1≤βγ ≤1000 the mean energy loss by a charged particles in intermediate Z materials with an accuracy of a few percent. In Fig. 2.1, the mass stopping power for positive muon, µ+, in Cu is plotted as a function of βγ =p/Mc. Below the lower limit, βγ < 0.1, the velocity of the particle is comparable to the atomic electron "velocities", and above the upper limit, the radiative effects become relevant. Note that particles with βγ ∼3, the minimum value of the stopping power formula, are called "minimum-ionising particles" or by its acronym, mip’s. In practical cases, most relativistic particles (e.g., cosmic-ray muons) have mean energy-loss rates close to the minimum [3]. Muon momentum 1 10 100 Mass stopping power [MeV cm2/g] Lindhard- Scharff Bethe Radiative Radiative effects reach 1% Without δ Radiative losses βγ 0.001 0.01 0.1 1 10 100 1001010.1 1000 104105 [MeV/c] 100101 [GeV/c] 100101 [TeV/c] Minimum ionization Eµc Nuclear losses µ− µ +on Cu Anderson- Ziegler Figure 2.1.: Mass stopping power [3]. 28 2.1. Interaction of radiation with matter 2.1.2. Energy loss by electrons The two main processes observed when electrons (or positrons) lose their energy interacting with matter are: collisions with other electrons in the material and, on the other hand, Bremsstrahlung. It is important to understand that the electrons tracks become straggly, as their mass is small, and so the corresponding Bethe-Bloch formulas are given for the linear mean path, instead of for its actual straggled one. Therefore: dE dx tot =dE dx coll +dE dx rad (2.1) The energy lost by electronic collision is roughly given by the Bethe- Bloch formula in Eq. E.3, however, it should be modified because two reasons: the small mass of electrons (which means that can be notably scattered in the matter on the contrary of heavier charged particles) and the fact of dealing with collisions between indistinguishable particles. Taking into account these two reasons, the linear Bethe-Bloch formula leads to: dE dx coll =−KZ A 1 β2lnmeβ2c2γ2T 2I2+F(γ)(2.2) where T is the kinetic energy of the electron and the function F is a correction factor different for electron or positrons. The Bremsstrahlung energy-loss mechanism is created by the Coulomb field of the nuclei, resulting in the emission of braking-radiation. The linear specific loss through this radiative process is dE dx rad =−NEZ(Z+ 1)e2α m2 0c44ln 2E m0c2−4 3(2.3) This process becomes relevant at energies above a few ten’s of MeV where its contribution is greater than the collision process [33]. Since the energy loss by braking-radiation depends strongly on the absorbing material, an important parameter is defined: the critical energy, Ec, where the probability of the radiation loss equals the probability of 29 2. Scintillator detectors Figure 2.2.: Fractional energy loss per radiation length in lead as a function of the electron or positron energy [3]. collision loss , defined by Eqs. 2.2 and 2.3. Fig.2.2 shows the contributions of both components for lead: the critical energy is around 7 MeV. and so former becomes dominant at lower energies and the latter at higher energies. 2.1.3. Energy deposited by photons When a photon interacts with matter, the processes involved are dependent on the energy of the incident photon and on the atomic number (Z) of the material. The main interaction of X-rays and γ-rays with matter is basically due to the three processes that will be shortly described here (for a more detailed explanation, see, for instance, Refs. [33, 19, 3] ): •The photoelectric effect. •The Compton scattering (including Rayleigh scattering). •The pair production. Photoelectric effect This effect is the more likely for photons of low energies and consists on the total energy transfer from the photon to an 30 2.1. Interaction of radiation with matter atomic electron, that is then ejected from the shell of the absorbing atom, and so the energy transferred to the outgoing electron is: Ee=hν −φ(2.4) where Eeis the energy of the ejected electron, hνis the energy of the incident photon and φis the binding energy of the electron. It is worth bearing in mind that the ejected electron can have energy enough to produce further excitations or ionisations, in a chained way. Compton scattering The Compton scattering process is dominant at intermediate energies. The incident photon interacts elastically with an electron and the total energy and momenta are conserved after deflecting an angle θ. The energy of the deflected photon is dependent on such angle, hν0=hν 1 + hν m0c2(1 −cos θ)(2.5) The highest energy that can be deposited, corresponding to full backscatter, is called the Compton edge. For low scattering angles, the energy transferred approaches to zero, meaning that the gamma is not detected. The outgoing electron can go through secondary interactions, and the deflected photon can, either escape or undergo a new photoelectric or Compton interaction. Pair Production (PP) The pair production contribution starts when γray energies are larger that the sum of the rest masses of the lightest leptons (e−and e+), being this energy threshold ¯ 1.022 MeV. Such process involves the presence of a nucleus in order to conserve momentum and so the angles of the created leptons partly depend on the energy of the initial γ-ray. Once again, it is worth mentioning that the lepton pair will produce further interactions in its track, being remarkable the case of the positron, that will ends by anhilitation with one electron of the medium, creating so two back-to-back photons of 511 keV that can newly interact at random. In Fig. 2.3 is shown the total interaction cross section for two different materials as a function of the energy of the incident photon . The figures 31 2. Scintillator detectors to be more efficient when they are dispersive. Both the election of wrapping materials, as well as a good tight coverage, are crucial when looking for the best efficiency in the scintillation light collection that redounds to the higher energy resolution. The most common used wrapping material is PTFE (Teflon) because of its high light reflection and its use ease; the ever most efficient material is Enhanced Specular Reflector (ESR) from 3M, [36]) whose reflectance is higher than 98%(see Ref [37] for a comparative study). The hygroscopicity of the materials can degrade the properties of the crystals, affecting to their transparency, (the so-called yellowing) and worsening the transmission of the light created in the scintillation process. Therefore, an appropriated encapsulation must be made in cases of high hygroscopicity to avoid the absorption of moisture from the environment. CeBr3CeBr3crystal is highly hygroscopic and, therefore, the encapsulation was made by the manufacturer. The crystal was surrounded by a 0.5 mm of Teflon reflector, and a quartz window (nquarz = 1.54) was placed over the exit window of the crystal. The assembly was then encapsulated in an aluminium housing with 1.5 mm thick walls. CsI(Na) CsI(Na) crystal is less hygroscopic than CeBr3, and so, the encapsulation could be made by us in our USC laboratory. The wrapping of the crystal consisted in two foils of ESR. Black adhesive tape was wrapped around plus two 2 mm sheets of G-10, in order to tight the crystal to the walls of the aluminium housing (30mm x 30mm x 30mm). The input crystal face was sealed using an aluminium adhesive tape of 0.75 mm thickness from 3M. CeF3_L50 This crystal is non-hygroscopic. Similarly to CsI(Na), the CeF3_L50 crystal wrapping consisted of two foils of ESR and black adhesive tape, plus two sheets of G-10 in an aluminium package of 25 mm x 25 mm x 50 mm. 38 2.3. Scintillator detectors used in this work. (a) (b) (c) (d) Figure 2.5.: Scintillator crystals used in this work: (a) CeBr3crystal encapsulated by Scionix. (b) CsI(Na) crystal. (c) CeF3crystal L = 50 mm. Note the non-polished surface. (d) CeF3crystal L = 40 mm wrapped by two foils of ESR from 3M reflector material. 39 2. Scintillator detectors CeF3_L40 CeF3_L40 was simply wrapped using two foils of ESR without aluminium housing. Depending on the sort of measurements, two wrapping configuration were performed: •A: having an opening of 2 mm in the input surface to let the α particles interact with the crystal. •B: a whole wrapping of two foils of ESR. 2.3.2. Photodetectors used Photodetectors can be classified mainly in two groups: Photomultiplier tubes (PMT) and Si-based photodetectors, as PIN diode (PD), Avalanche photodiode and Silicon PMT (SiPM). The photomultiplier tubes and PIN diode photodetectors will be briefly described in the following. For SiPMs see Ref. [38]. R7600U-200 Photomultiplier Tube The advances on PMTs in 90s decade had led to improvements in their design and development. An step forward was done by Hamamatsu at the R&D for the Super-Kamiokande experiment, boosting the technology applied to them. For instance, the noise level was reduced by interposition of insulator materials between the adjacent dynodes, preventing ions and light feeding-back to the photocathode in the electron multiplication process. The positions of the first and second dynodes were optimised for a good efficiency collection. In order to focus photoelectrons by electron-lens effect, more efficiently guiding photoelectrons from the photocathode to the first dynode, a new technology to arrange grid pattern between the photocathode and immediately in front of the first dynode was developed. The PMT R7600U-200 manufactured by Hamamatsu, besides the properties above mentioned, has also improved the feedback prevention by using metal packages (technology developed in 1992). In addition, a novel technology for optimising the position between the first and second dynodes has been used; this technology was invented for metal channel dynodes with no grid patterns (see Fig. 2.6) [4]. As a result of this inheritance, Hamamatsu developed the smallest TO-8 metal package photomultiplier 40 2.3. Scintillator detectors used in this work. tubes. The cubic shape of such PMT facilitates stacking the photodetectors, when a big detector requires large segmentation [39]. (a) (b) Figure 2.6.: Cross section of metal channel dynodes[4] (a). PMT mod. R7600U-200 from Hamamatsu (b). The PMT R7600U-200 also offers a very high photocathode quantum efficiency, 43 %at 430 nm, called "ultra bialkali" (UBA). Former photocathodes, with high sensitivity, called "superbialkali" (SBA), deliver a lower quantum efficiency of 35%at 350 nm. Fig. 2.7 shows typical spectral response curves of the UBA, SBA and BA photocathodes. (a) (b) Figure 2.7.: Q.E. of UBA, SBA and BA photocathodes for CeBr3(a). Spectral response of PIN diode model S32004-08. Courtesy of Hamamatsu Photonics K.K. (b). Just to see the evolution of the photocathode efficiency, let us mention 41 2. Scintillator detectors that the Q.E. of the first PMT, built in 1930, was less than 1 %at peak wavelength and then, the research and development of the photocathode materials got improved the Q.E. up to 25%in the mid-1960s. These increments of the Q.E. becomes very relevant for those applications and experiments where the number of photons emitted by the photon source (scintillator crystal) is scarce [40]. PIN Diode S3204-08 The PIN diode was one of the first developed silicon, Si, photodetectors. Their inner structure roughly consists on the junction of three silicon layers (in order): a Si layer heavily doped with donors (n+ region), a Si intrisic piece and a Si layer heavily doped with acceptors (p+ region). This configuration is reverse-biased, forming a depleted region that contains the whole intrinsic piece and extending to both the p+ and n+ regions, forming a wide light detection area (see Fig. 2.8). PIN diodes are usually covered by a thin layer of highly transparent resin for an optimal coupling as well as preventing a possible mechanical damage of the first doped silicon layer. Solid-state devices are advantageous over PMTs when operating in the presence of a magnetic field, that disturbes the swarm of electrons created in the PMTs stages and do not in the PIN diode case [5]. As a drawback the Q.E. of the PIN diode decreases at the wavelengths of our interest, being so its overall efficiency much lower that in the PMT R7600U-200. The radiant response for PIN diode S3204-08 is shown in Fig. 2.7 (b). The PIN diode, model S3204-08 from Hamamatsu, has a large active area with a compact design, as can be seen in Fig. 2.8; and its power supply voltage for a gain of x10 is around 70 V, as specified in Table 2.3. PMT R7600U-200 PIN diode Bias Voltage(V) -800 70 Gain 10710 Table 2.3.: Photodetectors biasing. 42 2.3. Scintillator detectors used in this work. (a) (b) Figure 2.8.: Structure of a PIN diode, (a) [5]. PIN Diode S3204-08 from Hamamatsu (b). Courtesy of Hamamatsu Photonics K. K. [6] 2.3.3. Coupling scintillators to photodetectors The four crystals were coupled to two different photodetectors attending to their output window and their emission spectra. Both the L.Y. of an scintillator crystal and the Q.E. of the photodetectors, are light wavelength dependent. The result of the convolution of both curves, called Photoelectron Yield, P.Y., is calculated and plotted in Fig. 2.9 for the three crystals, CeBr3, CeF3and CsI(Na) coupled to a PMT R7600U-200. Note the high P.Y. for CeBr3. Figure 2.9.: Photon Yield of CeBr3, CsI(Na) and CeF3[7],[8], [9] 43 2. Scintillator detectors Final assemblies CeBr3-PMT The R7600U-200 PMT and the CeBr3were coupled using two coupling materials: a Silicon Pad, mod. BC-634A, and optical grease, BC-630, both from Saint-Gobain Crystals [41]. The results obtained with the optical grease were comparable to those obtained when coupling with Silicon Pad. The set assembly will be called CeBr3-PMT onwards. CsI(Na)-PMT An aluminium piece was machined in order to a proper assembly between the crystal encapsulated and the surface of the R7600U- 200 PMT. This piece was glued to the plastic coating of the PMT. A ESR mask surrounding the borosilicate window that is set in order to avoid light losses. The coupling is made using a bi-component adhesive Scionix RTV 681 from Scionix [42]. The set assembly will be called CsI(Na)-PMT onwards. CeF3_L40-PMT and CeF3_L50-PMT Both crystals are coupled to the R7600U-200 using optical grease, BC-630 and an ESR mask as the decribed previously [41]. The sets assembly will be called CeF3_L40-PMT and CeF3_L50-PMT respectively onwards. CsI(Na)-PIN The CsI(Na) crystal is glued to the PIN diode S3204-08 using Meltmount n = 1.582 adhesive [43]. The set assembly will be called CsI(Na)-PIN onwards. 2.4. Simulation work Simulations are essential to understand the response of detectors to different kind of particles in function of its kinetic energy. Therefore, the simulations performed for two different type of crystals are described here in order to better understand the response of the scintillator crystals, for both γ-rays and electrons. Finally the results and conclusions will be presented. The features of the simulation program here used, EnsarRoot [17], are described in the Appendix F. 44 2.4. Simulation work 2.4.1. Description of the simulation Two punctual radiation sources -one for γ-rays, one for electrons, e−- have been simulated. The sources were placed at 1 cm from the input window of the crystals. Their emission angles are 0o≤θ≤15oand 0o≤ϕ≤ 360o. The simulation medium is vacuum. The software provides an output ROOT file with different observables. The most relevant parameters for this study are the deposited energy in the crystals for different incoming energies in order to find out the efficiency of the crystals for a certain energy of the γ-rays and the most probable value of the deposited energy in the electrons case. This simulation work has been performed for both CeBr3and CsI(Na) scintillators, with dimensions close to the ones having the real detectors used in the rest of the Thesis work. For the sake of simplicity, CsI(Na), the material simulated was CsI. The aim of the simulation performed in this work is to determine the energy deposited in each crystal and, therefore, the dopant, (Na in this particular case) does not alter significantly the characteristics of the CsI. The characteristics not affected by the dopant are the density of the material and its radiation length. See Table 10.9 in Ref. [44]. Nevertheless, the optical characteristics, such as wavelength emission or refractive index were significantly modified. Three different configurations were defined for each crystal as follows: a) Bare crystals. Dimensions (close to the real ones): 18x18x25 mm3. b) Crystals in contact with an Al foil. 1.5 mm thick and an area of 18x18 mm2. c) The same as in the configuration c) but with a foil thickness of 5 mm. The simulation results leads to the spectrum of deposited energy in the crystal as well as to an estimation of the counting efficiency. In order to get more realistic results an experimental smearing is included for each crystal, corresponding to the global detector resolution at 1 MeV.. 45 2. Scintillator detectors Figure 2.10.: Schematic set-up for simulations. The simulated source is sketched by the circle placed a t the origin of coordinates. The parallelepiped crystal is shown as well in the figure. The dashline represents the solid angle from the source to the end of the crystal. 46 2.4. Simulation work 2.4.2. Photon simulations The photon simulation consists on the emission of monochromatic γ-rays in a wide range of initial energies: from a few MeV to GeV. The simulation output provides a ROOT Tree that contains a set of observables, stored in a event-by-event basis. The observable of interest for the γ-ray simulation is the deposited energy by every incident γ-ray recovered from the Tree, using a script runs in ROOT. The same range of incident energies is simulated for the three configurations, stated above, of both crystals. The simulations were performed over 15000 events (15000 incident γ-rays). A study of the detection efficiency of the CeBr3and CsI was made. The detection efficiency relates the ratio of number of events registered by the crystal having an amplitude above a threshold well above a tipical SNR >1, over the events impinging on the crystal. In Fig. 2.11, (a) and (b), the respective values of such efficiencies for each incident energy and each configuration are shown. It is observed that the detection efficiencies rising up to 60%from a few MeV to GeV, where the efficiency saturates, because (see the spectra in Appendix G) the most probably deposited energy in these crystals never exceeds 20 MeV. A similar behaviour is observed in the case of the CsI crystal, although the efficiency values are slightly below than the ones in the CeBr3for low energies and slightly above at high energies. It is worth pointing out that at low energies, the photopeak efficiency, (the efficiency of the crystal when the the full-energy of the incident particle is registered by the crystal), decreases as the incident energy increases, (see Fig. 2.14) as it was observed in γ-rays radioactive sources spectra experimentally acquired in Chapter 3. It is clear that the size of the crystal plays an important role when dealing with the photopeak efficiency: larger crystals have higher photopeak efficiency because less energy escapes from the crystal. In this particular case, the crystals can be classified as small detectors in agreement with the mean free path of the γ-rays in the material [33]. From Fig. 2.11 it becomes clear how for 2 MeV the photopeak and Compton energies are the most probable, whereas at 5-10 MeV the escaping peaks becomes very relevant. As a conclusion, both crystals behaves very similarly and the effect of 47 2. Scintillator detectors 2.5. Conclusions to the Chapter 2 After reviewing the mechanisms of creating a certain light yield in scintillation crystals and the way this light is converted to an electronic pulse by the corresponding optodevice, are presented the simulation work done to better explain the results gotten both in lab bench tests and in the CERN-n_TOF experimental facility. The simulations performed using the EnsarRoot simulation program for CeBr3and CsI crystals have the clues to interpret provided the behaviour of both crystal for γ-rays and β particles. It is worth mentioning the high detection efficiency for γ-rays, greater than 40 % in the energy range studied and increasing up to 60 % at energies larger than 100 MeV. An simulation for the deposited energy by βparticles in both crystals provides accurate information about the threshold energy when radiation losses become relevant. 54 3. Tests result with radioactive sources and cosmic-rays In the previous Chapter those scintillator detectors studied in this work have been introduced and the challenges of dealing with fast and high- L.Y. scintillator crystals, as well as some technical solutions to face them were discussed in Chapter 1. In this Chapter is shown how the application of such techniques permits a proper characterisation and a better performance of these fast radiation detectors. Tests with radioactive sources were performed in order to characterise the response to the deposited energy, so that the detectors linearity, energy resolution, timing resolution and efficiency will be discussed. 3.1. Detectors calibration: spectra acquisition and linearity The radioactive sources using in the following sections in this chapter are mainly γemitters, namely 152Eu, 137Cs, 60Co and 88Y, but also a βemitter, 90Sr, an αsource, 238Pu, and finally, one of Am-Be, giving both γand neutron source were used. The tests done with these radioactive source allows a study of the detectors behaviour in a range of energies from 121 to 4444 keV. 3.1.1. Energy resolution and linearity The typical set-up to get the different spectra was described in Section 1.1.3 and the procedures there described have been used here. The low-noise high-stability HV power supply from ISEG, module NHQ 225M [46], was feeding the optodevices and the signal output was sent to a preamplifier 55 3. Tests result with radioactive sources and cosmic-rays and shaping stage (either the stretcher PCBs described in the Appendix C, or the MPR-1 module from Mesytec [47] in the case of CsI(Na)-PIN), then connected to a SA, CANBERRA mod. 2022, and, finally, the amplified and shaped signals were acquired by a MultiChannel Analyser MCA-8000A from AmpTek [48], that digitises and transfers the data to a PC where the client software have been installed. The spectra acquired from 88Yand AmBe radioactive sources were performed by the CeBr3-PMT and CsI(Na)-PMT detectors and both sources were placed at the EAR1 in the CERN-n_TOF facility. In this case, the set-up differed slightly from the measurements done at the USC laboratory with the chain above mentioned: the stretcher PCBs used were an early versions of the final designs, and the SA employed was an ORTEC mod. 570, from the shelves in the n_TOF experimental facility. The 152Eu source was used for the calibration of the CeBr3-PMT, CsI(Na)- PMT and CsI(Na)-PIN detectors because it features eight peaks of relatively high intensities. Their energy (in keV) and BRs are, respectively: 121 (25.6%), 244 (7.6%), 344 (26.5%), 778 (12.9%), 964 (14.6%), 1085 (10.2%), 1111 (13.6%) and 1408 (21.0%). The resolution of the photopeaks (see in Fig.3.1) were calculated, for every scintillator detector, from the parameters given by fitting to a gaussian function plus a linear function [49] after subtracting the background radiation. It is worth saying that the two peaks at 1.086 and 1.112 MeV are 26 keV apart from each other, what is around 2.36% of the peaks’ energy and, therefore, it can not be resolved. Nevertheless, its BRs are quite similar and so, in this study, the apparent peak shape has been unfolded in two gaussians, getting in this way the resolution attainable at its middle point, 1.1 MeV. 3.1.1.1. CeBr3-PMT In Fig. 3.2 (a), the channel-energy linear calibration is shown when used the stretcher PCBs originally designed for this work. In Fig. (b), the energy resolution obtained for the photopeaks of 152Eu were fitted to R(E) = p0 Ep1(3.1) 56 3.1. Detectors calibration: spectra acquisition and linearity Figure 3.1.: The 152Eu spectrum (in this case with the CeBr3-PMT detector). The energy resolutions given here are the result of the intrinsic resolution of the scintillating process in the detector, plus the gaussian dispersion of the L.Y., plus the dispersion of the multiplication by the opto-device, plus the contribution from the electronic chain. In order to quantify this electronic chain contribution, the signal output from the CeBr3-PMT was simulated, by an artificial pulse generator connected to the rest of the spectroscopic chain, including the stretcher and the SA. The output from a signal generator was shaped, by differentiation of its squared pulse by means of a RC circuit. The amplitude of this signal was arranged to be similar to the signal produced by a γ-ray of 662 keV; The result seen in Fig. 3.3 (b), showing a sigma =4 (FWHM =9.4), is leading to a Rpulser = 1.4 %. This dispersion includes the stability of the pulse amplitude from the signal generator, that was calculated by means of a 5 GS/s digital oscilloscope to have a σ/amplitude =0.5 %. We concluded, therefore, that the resolution factor introduced by the electronic chain was Relec = 1.3 %±0.2 %and the overall resolution was then unfolded by means of R2 detector =R2 total −R2 elec (3.2) A spectrum of 88Y+AmBe was measured. Fig 3.4 shows the efficiency 57 3. Tests result with radioactive sources and cosmic-rays (a) (b) Figure 3.2.: Calibration energy (left) and energy resolution (right) for CeBr3- PMT (a) (b) Figure 3.3.: 137Cs+60Co spectra using CeBr3-PMT detector (a). Electronic contribution resolution (b). of the detector to detect γ-rays of 4444 keV. The efficiency for such as high energies is low because of the small size of the crystals. Indeed, as it was shown by the simulations performed in this work (see Section 2.4 in the previous Chapter): this low photopeak efficiency is due to the high probability of secondary photons escaping by the crystal boundaries (see Section 2.1). In some published works, Refs. [50] [7], on crystals of CeBr3, inner background have been measured because the presence of contaminants 58 3.1. Detectors calibration: spectra acquisition and linearity in the material that could disturb the measurement of γ-rays spectra. The background acquired in this work along 24 hours, see Fig. 3.5, does not show the presence of contaminants in the CeBr3scintillator crystal provided by Scionix. Test with a βemitter A90Sr source was used to find out the ability of the CeBr3-PMT to detect βparticles. The spectra is shown in Fig. 3.6. Two contribution of the continuous spectrum can be distinguished, in spite of its aluminium housing: one is coming from the emission of the β−disintegration of 90Sr whose endpoint energy is 546 keV and the other is coming from the β−disintegration of the 90Y, whose endpoint energy is 2284 keV, being thus energetic enough to be able to pass through the aluminium sheet and to deposit energy in the crystal. Figure 3.4.: 88Y+AmBe spectrum acquired using CeBr3-PMT detector. 3.1.1.2. CsINa Similarly to CeBr3-PMT, the linear energy-channel calibration as well as the energy resolution depending on the energy (see Fig. 3.7. were calculated for CsI(Na)-PMT using the 152Eu radioactive source. In Fig. 3.8 is observed the ability of the CsI(Na)-PMT to detect γrays from 662 to 4444 keV with good resolution, although the detection 59 3. Tests result with radioactive sources and cosmic-rays Figure 3.5.: Background radiation spectrum measured using CeBr3-PMT detector. No intrinsic activity is observed. Figure 3.6.: 90Sr spectrum. Note the two contribution of the desintegration branches at 546 keV and 2284 keV, after passing the Al wall. 60 3.1. Detectors calibration: spectra acquisition and linearity efficiency becomes lower at the higher energies. 61 3. Tests result with radioactive sources and cosmic-rays (a) (b) Figure 3.7.: Calibration energy (left) and energy resolution (right) for CsI(Na)-PMT 62 3.1. Detectors calibration: spectra acquisition and linearity (a) (b) Figure 3.8.: CsI(Na)-PMT 137Cs+60Co (a) and 88Y+AmBe spectra (b). 63 3. Tests result with radioactive sources and cosmic-rays (a) (b) Figure 3.14.: Temperature variation over the time for CsI(Na)-PIN detector. Figure 3.15.: Resolution variation over the temperature for CsI(Na)-PIN detector. 70 3.1. Detectors calibration: spectra acquisition and linearity Temperature dependence for CeBr3-PMT The method to measure the temperature variations was the same than described for the CsI(Na)- PIN. The temperature range measured in this case, can be seen in Fig. 3.16 (a) (b) Figure 3.16.: Temperature variation over the time for CeBr3-PMT detector. Figure 3.17.: Resolution variation over the temperature for CeBr3-PMT detector. As it was shown in the preoviuos measurments, for the CeBr3-PMT the temperature do not greatly affect the resolution values obtained for the photpeak of 662 keV γ-rays. This agrees with Fig. 9 in Ref [54], where the deviation of the L.Y. is around 1 %in the range of temperatures studied. 71 3. Tests result with radioactive sources and cosmic-rays 3.2. Pulse Shape Analysis The shape of the pulses generated by either the ionising particles or the γ-rays is an brings out some information about its kind. The signals produced by the interaction of different ionising particles γ-rays, βparticles and cosmic-muons will be analysed in this Section using the techniques described in Chapter 1 3.2.1. PSA through digital algorithms The signals provided by the scintillator detectors, CeBr3-PMT and CsI(Na)- PMT, were acquired and digitised by the oscilloscope YOKOGAWA DL9140 (Max. sampling rate 5Gs/s and Freq. BW = 1GHz) and transferred to a PC for the analysis explained in the following sections. 3.2.1.1. Raw signal analysis The decay time of each detector was firstly calculated by a fitting method; secondly the shapes of the signals produced by the three different particles mentioned previously were compared; and finally, the timing resolution of the detectors was obtained by the coincidence with an ultra-fast organic scintillator detectors. Decay time constant In order to obtain the decay time for each detector, the profile of a set of signals haven been calculated. The profile consists on the calculation of the mean value of each same sample of the many signals acquired. In this case, the signals were produced by the 622 keV γ-rays from a 137Cs source. The decay time constants were calculated from a exponential fitting (as shown if Eq. 2.8) of the decay time of the profile. Figs. 3.18 (a) and (b), show the fitting for CeBr3and CsI(Na) respectively. The τCeBr3is 12 ns, in agreement with the value provided by the references in Table 2.1. When fitting the CsI(Na) profiles two decay time constants have been found : τ1,CsI(Na)= 473 ns and τ2,CsI(Na)= 2588 ns. The first constant differs from the provided in Ref. [9], where the value is 630 ns. But it is excelent agreement with Ref. [55] where a decay 72 3.2. Pulse Shape Analysis constant equals to 472 ns have been reported. The result of 473 ns found in this work rules out the value of 630 ns given by Saint-Gobain Crystals [9]. / ndf 2 χ 0.5303 / 354 Constant 0.2164± 0.4194 Slope 0.01013± -0.08219 time (ns) -10 0 10 20 30 40 50 60 70 Amplitude (arbitrary units) -3 10 -2 10 -1 10 1 / ndf 2 χ 0.5303 / 354 Constant 0.2164± 0.4194 Slope 0.01013± -0.08219 = 12 ns 3 CeBr τ (a) / ndf 2 χ 0.1967 / 1765 p0 0.3233± 0.8092 p1 0.0001036± -0.0003864 p2 0.068± 2.391 p3 0.000297± -0.002113 Time (ns) -500 0 500 1000 1500 2000 2500 3000 3500 Amplitude (arbitrary units) 0 2 4 6 8 10 / ndf 2 χ 0.1967 / 1765 p0 0.3233± 0.8092 p1 0.0001036± -0.0003864 p2 0.068± 2.391 p3 0.000297± -0.002113 = 473 ns 1 τ = 2588 ns 2 τ (b) Figure 3.18.: CeBr3and CsI(Na) decay times. 73 3. Tests result with radioactive sources and cosmic-rays Pulse shape for γ-rays, βparticles and cosmic-µ’s The linearity is relevant when studying both the timing and energy resolution properties of the detectors. The profile of a set of signals from three different ionisation particles (γ-rays from 137Cs,βparticles from 90Sr and cosmic-µ’s) has been calculated in order to find out if their respective pulse shape show variations. The setup for measuring the cosmic-µ’s is detailed described in Section 3.2.1.1 In Figs. 3.19 and 3.20 the profiles for each ionisation particle has been normalised to the integral of the pulse. It is is remarkable that, both for CeBr3-PMT and CsI(Na)-PMT, the three pulse shapes are similar without any noticeable deviation. It worth mentioning that mean maximum amplitude of the cosmic-rays µ’s is around 10 times larger than the case of the corresponding to the 662 kev 137Cs, concluding the noteworthy linearity of both detectors. Figure 3.19.: CeBr3-PMT profiles for γ-rays, βparticles and cosmic-rays. Timing resolution A plastic model EJ-232 from ELJEN TECHNOLOGY, wrapped with a foil of mylar and black adhesive tape, coupled to a PMT mod. R4998 by means of a Silicon Interface Pad. BC-634A, manufactured by Saint-Gobain Crystal, was placed in parallel with the scintillator detector. The signals produced both in plastic detector and in the scintil- 74 3.2. Pulse Shape Analysis Figure 3.20.: CsI(Na)-PMT profiles for γ-rays, βparticles and cosmic-rays. lators detectors were registered in the oscilloscope YOKOGAWA DL9140 using a coincidence trigger. Most of these registered events are cosmic rays muons, the most abundant charged particles at sea level, whose mean energy corresponds to ≈4 GeV [3]. The incident angle of the majority of muons at sea level is close to the zero zenith angle and their flux is ∼1 muon/cm2minute. The time resolution (σ) was is obtained by the characterisation of the statistical distribution of (tCeBr3-tP lastic), where tCeBr3is time obtained by the intersection of the linear extrapolation of the rising edge with the baseline. The tPlastic is the time when the signal crossed a constant threshold. The rising time of the plastic scintillator is shorter than 4 samples of the digital signal. The distribution was fitted with a gaussian function. As shown in Fig. 3.21 the timing resolution (σ) for the CeBr3-PMT is 0.230 ns. 75 3. Tests result with radioactive sources and cosmic-rays / ndf 2 χ 22.76 / 15 Constant 7.3± 173.3 Mean 0.0074± 0.4225 Sigma 0.0062± 0.2291 Tcrys-Tplast (ns) -2 -1 0 1 2 3 4 Counts 0 20 40 60 80 100 120 140 160 180 / ndf 2 χ 22.76 / 15 Constant 7.3± 173.3 Mean 0.0074± 0.4225 Sigma 0.0062± 0.2291 Figure 3.21.: CeBr3-PMT timing resolution distribution. 76 3.2. Pulse Shape Analysis 3.2.1.2. Using Triangular and Trapezoidal filters Triangular smooth The registered output signals from CsI(Na)-PMT, produced by γ-rays from 137Cs and 60Co sources, shows an inaccurate shape that hinder the accuracy when defining the observables. To improve such inconvenient, a smooth filter is passed over the original signals. A triangular filter is chosen an its number of parameters is 5. The filter is passed twice over the signal. In Fig. 3.22 the improvement in the shape for one signal can be appreciated. Fig. 3.23, the rise region is enlarged to show the improvement in the shape of the original signal after one and two pass of the smooth filter. Figure 3.22.: Original pulse from CsI(Na)-PMT (black) and smoothed pulse (orange). Once the signals have been conditioned, the same procedure that was detailed for the CeBr3-PMT detector is used to calculate de CsI(Na)-PMT timing resolution. The timing distribution is shown in Fig.3.24 where the timing resolution (σ) 0.920 ns for the CsI(Na)-PMT. The observable pulse area have been calculated for the set of pulses. This observable contained information about the deposited energy of the γ-rays emitted from the 137Cs and 60Co. In Fig. 3.25 the distribution of this observable can be seen. The three photopeaks emitted by the sources, 662 keV, 1173 keV and 1332 keV can be distinguished. The peak at 662 77 3. Tests result with radioactive sources and cosmic-rays time (ns) -20 0 20 40 60 80 100 120 Amplitude (V) 0.00 0.05 0.10 0.15 Raw signal time(ns) -20 0 20 40 60 80 100 120 Amplitude (arb. units.) 0.00 0.05 0.10 0.15 Triangular smooth. n = 5, 1 pass time(ns) -40 -20 0 20 40 60 80 100 120 Amplitude (arb. units) 0.00 0.05 0.10 0.15 Triangular smooth. n = 5, 2 pass Figure 3.23.: Rising time for a CsI(Na)-PMT pulse. Upper Fig. Original signal. Middle Fig. Signal after one pass 5th order triangular smooth. Lower Fig. Signal afer two pass 5th order triangular smooth. / ndf 2 χ 9.402 / 3 Constant 11.0± 217.5 Mean 0.037± 3.039 Sigma 0.0389± 0.9191 (ns) plastic -T CsI(Na) T -4 -2 0 2 4 6 8 10 Counts 0 50 100 150 200 250 / ndf 2 χ 9.402 / 3 Constant 11.0± 217.5 Mean 0.037± 3.039 Sigma 0.0389± 0.9191 Figure 3.24.: CsI(Na)-PMT timing resolution distribution. 78 3.2. Pulse Shape Analysis keV was fitted using a Gaussian function in order to obtained a energy resolution value of such spectrum. The resolution obtained is R(%) = 6.7 ±0.1, which is in agreement with the resolution value measured using the spectroscopic chain in Section 3.1.1.2. Figure 3.25.: 137Cs and 60Co spectrum obtained after a smoothing of CsI(Na)-PMT signals. The energy resolution for 662 keV R(%) = 6.7±0.1 Trapezoidal filter The trapezoidal filter was applied to the digitised points of the pulses produced by the CeBr3-PMT. In Fig. 3.26 the original signal and the filtered can be seen. The parameter chosen was L = 12 ns, G = 4 ns. It is noticeable the reduction of the SNR that simplifies the calculation of the observables such as time stamp, pulse height or area. Figs. 3.27 shows a 137Cs and 60Co pulse height spectra for CeBr3-PMT. The peak was fitted to a Gaussian function, giving as a result an measured energy resolution R(%) = 8.2 %0.1, a value poorer than in the measured in the case of the spectroscpic chain case. See Section 3.1.1.1. 79 3. Tests result with radioactive sources and cosmic-rays 86 4. Beam-tests in the n_TOF experiment The significantly high flux of neutrons attainable from the high-intensity short-pulsed 20 GeV protons on the nTOF spallation target, together with the very special characteristics of the two lines of neutron time-of-flight, mades the n_TOF measurements superseding those at any other neutron source world around. These kind of facilities are devoted to enlarge and improve the existing libraries on nuclear data, that are foundations of Nuclear Science, because the knowledge of nuclear evaluated data is essential for the deeper understanding of many of the processes occurring in Nature. In particular, for disentangling the puzzle of the nucleosynthesis of elements beyond Iron in stars, via the s- and r-processes [57] and on the other hand, for boosting the development of technologies related to the better functioning of present nuclear reactors [58] as well as the future ones (Generation-IV) aiming at minimizing the production of nuclear waste or at burning it [59] in Fast reactors, namely in ADS [60]. This makes it necessary to perform new and more accurate measurements of many isotopes and reactions, which are summarized for instance by the Nuclear Energy Agency in its High Priority Request List [61]. In Chapter 3, the detectors have been characterised in a lab bench, where the measurement conditions can be largely controlled, whereas in an experimental area devoted to the study of nuclear reactions the environment becomes challenging when performing the measurements. The scintillator detectors previously characterised were moved to the n_TOF experiment facility at CERN in Geneva (Switzerland), in order to test their behaviour in a real environment, providing new information for future possible use in new detectors. This chapter deals with the measurements related to the n_TOF facility at CERN) done in the frame of this Thesis work. The detectors CeBr3-PMT and CsI(Na)-PMT were firstly tested in the n_TOF facility by measuring the prompt γ-ray emission in the neutron capture reactions 87 4. Beam-tests in the n_TOF experiment (n,γ) for the isotopes 197Au and natAg. In addition to this, the prompt-γ produced in neutron-induced fission 235U(n,f) were measured in coincidence with the PPAC gaseous chambers. Then, after a brief description of the n_TOF facility, the measurements and results obtained on the study of its γ-flash are shown in Section 4.4.1. 4.1. n_TOF Facility The neutron time-of-flight facility n_TOF features a white neutron-source produced after nuclear spallation reactions in the hadronic cascades originated in a huge lead target by protons bunches of 7 ns (rms) from the Proton Synchrotron (PS) of 20 GeV/c, having 7x1010ppp, at a repeating rate of 0.8 Hertz. The facility, aiming primarily at the measurement of neutron-induced reaction cross sections, was operative at CERN between 2001 and 2004, and then underwent a major upgrade in 2008 (See Ref. [62] and references therein). The required high energy-resolution is attained by making the pulsed neutron-beam, that spans over an available energy range wide of ten decades, to travel a distance long of 185 m before reaching the sample under study. In this technique the kinetic energy of the neutrons is so determined from their time of arrival at the measuring station, ranging from some 600 ns, for those travelling close to the speed of light, up to around 80 ms for the thermal ones. The lead spallation target, whose dimensions are 40 cm long and 60 cm in diameter, is sketched in Fig. 4.2. Two different water layers are besides the target: one of 1 cm of distillated water acting as cooler and another of borated water, 4 cm wide, that works as moderator of the initial fast neutrons to enhance the low energy part of the spectrum. A wider and flatter energy spectrum, from thermal energy to relativistic neutrons, was so obtained (see Fig 4.3). During the spallation process, besides neutrons, a huge quantity of particles are produced due to both electromagnetic and hadronic processes. Therefore, the neutron beam line is tilted 10orespect to the proton beam, in order to suppress the presence of some of the produced neutral particles such as π0,K0and Λ, as well as reducing the g-flash. 88 4.1. n_TOF Facility Figure 4.1.: n_TOF facility Figure 4.2.: Lead spallation target. 89 4. Beam-tests in the n_TOF experiment Figure 4.3.: Evaluated neutron flux for different spallation targets in the EAR1 [10]. To sum it all up, the n_TOF facility provides a very high instantaneous intensity of around 107neutrons/pulse in a collimated beam of 3.5 or 10 cm, and an excellent energy resolution in the complete energy interval of interest. Due to these characteristics, the n_TOF facility stands today among the world leading facilities for measuring neutron-induced reactions. Beam line to EAR1 Figure 4.4.: Beam line EAR1. This beam line was done taking advantage of a service tunnel already existing at CERN. It goes from the spallation target to the Experimental 90 4.1. n_TOF Facility Area 1 (EAR1), having a length of 182.3 meters with different elements that are placed all along the beam line, as sketched in Fig. 4.4. In order to clean-up the neutron beam a 3.6 Tm sweeping magnet (200 cm in length, 44 cm gap) at 145.4 m, deflects the charged particles that are in the beam. After this magnet, at 151 m, a 3.2 m thick iron wall plays a role as µshielding because of the in-flight decay products (mainly µs and νs) from the K±s and π±s coming from the spallation process. The coarse shielding at around 80 m and two collimators are included in the beam-line in order to get a well defined neutron beam, as clean of halo as possible. The first collimator is at 136.7 m and the second one at 178 m. Two configurations for the second collimator can be chosen, depending on the measurement requirements: a 18 mm diameter with 235 cm of steel plus 50 cm of borated polyethylene is chosen to measure capture reactions, and a 80 mm diameter, with 50 cm of borated polyethylene, plus 125 cm of steel, plus further 75 cm of borated polyethylene, to measure fission reactions. A 3 m width broad concrete wall separates the previous elements from the EAR1. Finally, a 1.6 m deep concrete wall at 190 m separates the EAR1 from the DAQ area. where the experimental set-up is deployed; the beam line prolongs some 10 m from the end of the EAR1, performing a scape-line ending in a borated polyethylene cube at 200 m, to dump the beam [62] in order to minimise the back rebounding neutrons [63]. Beam line to EAR2 In summer 2014, a new neutron beam line was added to the n_TOF facility, going to the Experimental Area 2 (EAR2). It is vertically oriented, starting on top of the spallation target, which is there surrounded by only the layer of 1 cm of de-mineralised water acting as coolant.The entrance window is made of an Al foil coupled to the vacuum pipe flange. Different elements are distinguished along the beam pipe, at distances taken from the center of the spallation target. The first collimator at 7.4 m above the target, consisting of a Fe cylinder 1 m in length with 20 mm of inner diameter. At 10.4m a permanent dipole magnet is installed to deflect the charged particles produced during the spallation process. The magnetic field created is 0.253(4)T in the center of the magnet, whose length is 1.134 m. The integrated field of the magnet is 0.287 Tm. The 91 4. Beam-tests in the n_TOF experiment Figure 4.5.: Beam line EAR2. 92 4.1. n_TOF Facility dipole magnet inner gap is 340 mm, where the vacuum pipe is installed. A second collimator is installed at 25.04 m inside a large vacuum vessel with 680 mm outer diameter. The collimator consists of 2 m Fe and 1 m borated polyethylene (B-PE), where a core of boron-carbide (B-C) cylinders are inserted in the last 0.4m. The diameter of the collimator is variable following a conical shape, decreasing as the distance is closer to the experimental area, from 70.0 to 21.8 mm. The exit of the second collimator is situated at 120 mm below the floor of the EAR2. The focal point, which is theoretically the position of the narrowest spatial beam profile, is positioned at 1.08 m from the floor of the EAR2. The floor of the EAR2 is located at 18.16 m while ceiling is at 23.66 . The first meter from the floor is devoted to a neutron flux monitor, which is installed inside the vacuum pipes in the lowest section of the EAR2 and a Pb shielding. The different detection systems can be placed thanks to a set of support structures which are mounted on a pillar system the exit of the neutron beam. Finally, the beam dump is hanging from the roof of the building, being designed to respect the limits of dose rate of radiation protection for low-occupancy areas, 2.5 µSv/h, outside the building. It consists of three layer of B-PE, B-Fe core+ Fe blocks and concrete to fully the neutron beam. Fig. 4.5 sketched the beam line 2. See ref [64] for further details. In order to minimised the background, the following considerations have been taked into account: 1. From 0.9 m above from the spallation target up to 7 m concrete elements are placed. 2. 4 tons of stainless steel spheres formed the shielding in the middle part of the beam line, from 12.6 m to 18.4 m. 3. Plates of lithium-polyethylene surrounding the beam pipe are placed on the floor of the EAR2 to absorb thermalised neutrons. 4. 57 cm of Pb disks are placed after the exit of the second collimator inside the vacuum chambers to reduce the γbackground coming from the last section of the collimator. 5. Shielding elements such as Pb, B-PE and B4C were placed in the last vacuum beam -line, fixed to the roof of the EAR2, just before the beam dump in order to reduce the photons and neutrons streaming 93 4. Beam-tests in the n_TOF experiment back to the detectors. 4.1.1. Detectors at the n_TOF facility A wide variety of detectors are used at n_TOF for performing experiments to measure nuclear data from neutron-induced reactions as radiative capture, fission or light-nuclei spallation. Among these are: solid-state detectors, such as the silicon based neutron beam monitor [65] or CVD diamond detectors [66]; scintillation detectors, including a TAC made of BaF2crys- tals [13] and the C6D6 liquid scintillators [67] used for radiative capture detection; and, finally, gas-ionisation detectors, as the MicroMegas-based detectors [68], H19, a calibrated fission chamber from the Physikalisch Technische Bundesanstalt [69], or different sets of Parallel Plate Avalanche Counters, that are deeper described below [11, 12, 70]. Parallel Plate Avalanche Counters, PPAC As already commented, the studies performed in this work were done using the scintillating detectors in parallel with the PPAC detectors that allow identify the (n,f) reaction events. Let’s so to introduce the PPACs functioning and how they were used to our purposes. Parallel Plate Avalanche Counters (PPAC) belongs to the gaseous detectors type that operate in proportional mode after application of an electric field between two parallel plates surrounded by a gas at low pressure. The type of gas in combination with its pressure and the distance that separates the plates, play an important role when detecting the charged particles passing through the detector. In the proportional regime, the charged particle creates an specific number of pair electrons and positives ions that are drifted to the anode and the cathode respectively. During such drift, both the electrons and ions are able to create new pairs of electrons an ions. Such secondary electrons are accelerated enough due to the electric field that can ionise molecules and new electrons will be released creating an avalanche, form by a swarm of electrons that drifts to the anode, up to the gas moderate and extinguish it in a characteristic time given by the type of gas. The avalanche produced, 94 4.1. n_TOF Facility known as Townsed avalanche and proportional to the charge created in a first step following the Eq. 4.2 , induces charged both the anode an the cathode and as a result, a current signal is obtained where the electron contribution is much faster than the ions due to the fact that the mobility of the ions in a gas is around a 1% of the electrons. To a timing analysis the ions contribution will be filtered and remove of the output signal. dn n=α·dx (4.1) Where αis the firs Townsed coefficient. In the case that the electric field is uniform and constant, as in the case of parallel plate geometry, the coefficient αis a constant. Thus, the solution to the previous equation is: N(x) = N(0) ·eαx (4.2) The gas used in the PPAC setup is C3F8, a non-flammable gas. The PPAC configuration adopted is illustrated in Fig 4.6 based on the model proposed in ref. The active area of detection is a 20 cm x 20 cm surface made of 1.5µm Mylar material coated with aluminium on each side. Both cathodes have the same size, however, the mylar suface is not continuous but stripped with a 2 mm pitch: on horizontally and the other vertically what allows the tracking of the FF traversing the PPAC detector, see Fig. 4.6. A gap of 2.5mm is in between the anode and each cathode. The bias voltage applied to the is 300V/cm·mbar. Signal induced in both the anode and the cathodes are read and amplified by the associated electronics set in the plastic frame. In the case of the cathodes a delay line, connected to a preamplifier, is set at the end of the strips, being decoupled by capacitors of pF welded between the strips. The position of the avalanche is obtained by measuring the delay time between the pulse on the anode and the pulses propagating along the delay lines. The main advantage of the delay line is the reduction of readout channels: two per cathode instead of one per strip. PPACs, as a detector working on the proportional regime, are not efficient for the detection of γ-rays since the probability for the γradiation to interact in the gas is small. However, the detection efficiency for charged and FF fragments reaches up to 100%. 95 4. Beam-tests in the n_TOF experiment (a) (b) Figure 4.11.: Trapezoidal-filtered signals for CeBr3-PMT and CsI(Na)-PMT. In (a), the blue line corresponds to a couple of the events from CeBr3-PMT registered by the n_TOF DAQ. The red line is the filtered signal, showing a constant time offset. In (b), an original signal from CsI(Na)-PMT (black line) shows a baseline shift. The green line is the smoothed original signal showing a better shape. The red line corresponds to the trapezoidalfiltered signal. Note the baseline shift correction. 102 4.2. Prompt γ-rays from (n,γ) reactions Acquisition Sampling Rate 500 MS/s Scintillator detector CeBr3-PMT CsI(Na)-PMT L (ns) 16 560 G (ns) 4 56 Delay time (ns) 12 1160 Table 4.1.: Trapezoidal filter parameters for CeBr3-PMT and CsI(Na)-PMT detectors used in promt γ-rays mesurements at n_TOF. 4.2.2. Data analysis The capture spectrum has been calculated through the time difference of the γ-ray impact on the scintillator detector and the time of the γ-flash as a reference time. No PKUP correction has been carried out in this calculation. The neutron energy selected for this study ranges from 0 to 100 eV. To find the time-of-flight of the signal produced by the γ-rays a trapezoidal filter was used. These filters, one dedicated to the CeBr3-PMT signals, one dedicated to CsI(Na)-PMT signals, improve the signals making them higher, better shaped and baseline restored than the original ones. Two examples of filtered signals can be seen in Fig. 4.11. In (a), the blue line corresponds to a couple of the original signal from CeBr3-PMT with different amplitude registered by the n_TOF acquisition system and scaled for comparison. The red line is the filtered signal. In (b), a original signal from CsI(Na)-PMT (black line) shows a baseline shift. The green line is the smoothed original signal to show a better shape. The red line corresponds to the filtered signal, where the baseline shift has been corrected. As it is observed in both filters, a delay time is present in the filtered signals due to the filtered processes. Such effect is so relevant and must be corrected in order to obtained an accurate measurement of the time-of-flight. The filter parameters used in these cases are shown in Table 4.1. 4.2.3. Capture results The spectra obtained for the two samples and two detectors are shown in Figs. 4.12 and 4.13 where their shapes are compared with the ENDFBVII.1 evaluated library in the case of 197Au(n,γ) and CENDL-3.1 evaluated 103 4. Beam-tests in the n_TOF experiment (a) (b) Figure 4.12.: 197Au neutron capture cross section for 197Au. Red line: ENDF/B-VII.1 evaluation. Black line: measured scaled for comparison with the evaluation. (a) corresponds to CeBr3- PMT measurements, (b) corresponds to CsI(Na)-PMT measurements. 104 4.2. Prompt γ-rays from (n,γ) reactions (a) (b) Figure 4.13.: natAg neutron capture cross section for 197Au. Red line: CENDL-3.1 evaluation. Black line: measured scaled for comparison with the evaluation. (a) corresponds to CeBr3- PMT measurements, (b) corresponds to CsI(Na)-PMT measurements. 105 4. Beam-tests in the n_TOF experiment library in the case of natAg (n,γ). The energy spectra was corrected by the neutron flux for the 2011 campaing. Figure 4.14.: Measured counting rate distributions corresponding to the 197Au and the dedicated background measurements. Fig. taken from Ref. [13]. A background contribution is observed in every measurement at the lower region. As it is shown in Fig. 4.14 (see Ref. [13]), the contribution is mainly due to the fact of the activity of different materials in the experimental area, due to the in-beam γ-rays of of 2.2MeV from the 1H(n,γ) reaction produced in the moderator material surrounding the spallation target see Ref. [74]. In any case, the rate between the resonances below 5 eV and the background contribution shows that the neutron capture reaction contribution is greater than the background. It is worth noting the green line in Fig. 4.14, showing how the beam-off noise raises at the lower energies; this is because the lower energies correspond to the larger time bins. For the shake of comparison of the measured spectra of the 197Au and natAg fand the above mentioned evaluated libraries, it were normalised to 106 4.3. Prompt γrays from U5(n,f) reaction the maximum value obtained by the largest resonance. These capture measurements allow to tune-up the scintillator’s setup, and its results confirm that the radiation detectors and the analysis techniques used in this work, are suited to measure the prompt γ-rays emitted from (n,γ) reactions. 4.3. Prompt γrays from U5(n,f) reaction After the capture measurements, some tentative measurements were done on the 235U(n,f) reaction, where an important portion of the energy released, around 7%, is due to the prompt γ-rays. The request of a better characterisation of the fission γyields have been formulated in the Nuclear Data High Priority Request List of the Nuclear Energy Agency since the required accuracy, a 7.5% [75], was not achieved in previous measurements done in 1970’s [76, 77]. A better understanding of the fission process and a continuous development of nuclear technologies demand a detailed and precise study of the fission processes, mainly the yield of fission fragments, FF, and both the neutrons and the γrays emitted during the fission process toward stability. The deeper study of the neutron emission during the fission process encourage, however, the study of the prompt γ-rays that implies uncertainties on the nuclear data needed for the design of new fast reactors. For example, the local heating can be mispredicted by up to 30%even in well known power reactors [78]. 4.3.1. Experimental setup During the 2012 campaign, two scintillators detectors, already described in the previous Chapters, were placed in parallel to the PPAC detectors (above described in Section 4.1), using the configuration showed in Fig. 4.15 . The two scintillator detectors stood near PPAC 7 and PPAC 8, where one 235U target were in between, in order to get measured the prompt γ-rays resulting from de fission process, in coincidence with the PPAC pulses, that show to be faster than those from the scintillators, as 107 4. Beam-tests in the n_TOF experiment it is noticeable in Fig. 4.8. Thus, a method for coincidences, based on a specific trigger routine, has been developed. The main goal was observing the γ-rays from the resonances at low neutron energies. Figure 4.15.: PPAC targets Nine targets and ten PPACs are contained in the same tight enclosure and so the challenge was to identify the origin of each fission event, with its corresponding time stamp. The coincidences were found as follows: the γ-flash time was obtained as a initial reference for both the scintillators crystals and for the PPACs anodes labeled as 7 and 8. Two time windows were then defined in order to find out the signal coincidences between the anodes and the scintillator detectors. Two different studies were performed: on the one hand, the study of timing signal coincidences between the CeBr3-PMT and the PPAC anodes labelled and on the other hand, the study of the coincidences between the CsI(Na)-PMT and the anodes 7 and 8. 4.3.2. Data analysis Coincidence trigger Firstly, the coincidence trigger used for such purpose is here explained. In order to get the coincident measurements between both detectors, an algorithm was created that identifies all possible simultaneous signals. The main idea was to use binary arrays that can be identified with each PPAC and crystal detector. That simplifies later the analysis, chosing any coincidence condition just by applying a simple binary mask. For example, when wanted to identify coincident signals between PPAC #8 and the crystal the condition (mask) applied to the bit array would be 108 4.3. Prompt γrays from U5(n,f) reaction (a) (b) Figure 4.16.: Set up PPACs, scintillators 109 4. Beam-tests in the n_TOF experiment as follows: if( num & (1«8) == 1 ). The logical &bitwise AND operator checks for the value of the eighth bit stored in num, that is the bit array created to tag the signals and store them as an integer value. In case of multi-coincidence events, more complex structures can be required. For instance, the coincidence between the crystal, and the 7th and 8th PPAC detector can be expressed as follows: if (num & (1 << 8) == 1 && num & (1 << 7) == 1),(4.6) or even simpler: if (num & (1 << 8+1<< 7) == 1).(4.7) Furthermore, where anti-coincidence with other detectors are relevant, it can be easily selected by asking the specific bits to be explicitly zero. The code works as follows, first, the measured times and the trees with data from different detectors are synchronized and calibrated. Then, the algorithm loops over the sorted by time events and in case it finds two concurrent signals within a time window of 50 ns it stores the time, the corresponding neutron energy, the velocity and other useful information as the amplitude, area and width of the signal for further analysis. The identification number of the PPAC detector is stored into a binary list for that event as true. Then a subsequent loop is done over the remaining detectors in order to identify other possible positive signals. These coincident measurements are added to the binary list as true values as well, conforming the final value of num to be stored in the tree. In parallel, also all the coincidence signals between the PPAC detectors are stored into another binary array. The schematic algorithm design is shown in Fig.4.19. Analysis windows tppac is the time stamp when a pulse on the anode of the PPACs is detected. The routine used to analyse and store the data coming from the scintillator detector, routines used in Section for (n,γ) reactions 4.2, differs from the routine applied in the PPACs case as it was explained before. The coincidences are calculated as follows: 1. Searching coincidence between the crystal and one of the 10 anodes 110 4.3. Prompt γrays from U5(n,f) reaction is get by the condition: |tScin −tAnode|< timing window (ns)(4.8) where the timing window is different for CeBr3-PMT and CsI(Na)-PMT. Once such coincidence has been found, a flag is stored in the Tree in order to avoid a repetition of the coincidence in the second search of the coincidence, as was explained in above paragraph. 2. A new coincidence search between the time of the signal anode previously found in point 1, is carried out. The timing distance between both time stamps is 25 ns. |tAnode −tAnode0|<25 ns (4.9) A second flag indicates that the second coincidence has been found. 3. The parameters characterising the pulses, both the anodes of the PPACs and the scintillator detectors are stored in a new Tree. The timing windows were find out using an iterative method from the low time window related to the time width of the PPACs signals (FHWM) and increasing it until the background is appearing in the neutron energy spectrum. The timing distribution of time difference for the average time of anodes 7 and 6 for both CeBr3-PMT and CsI(Na)-PMT are shown in Fig.4.20. Both distribution shows non symmetrical timing windows. In addition, Figs. 4.17 shows this study for a set of timing windows taking into account the number of events registered as true coincidences. Two different contribution are observed. The flatter ones (black points) were fitted to a linear function and subtracted from the original plot giving as a result the plot formed by the blue dots. Considering the previously exposed, the timing window value chosen for the CeBr3-PMT case is a symmetrical window, |50|ns, and a non symmetrical from -100 to 300 ns in the CsI(Na)-PMT. 111 4. Beam-tests in the n_TOF experiment 4.4. γ-flash study 4.4.1. What γ-flash means? During the spallation process, when protons of 20 GeV/c impinges on the lead target, a huge amount of radiation and both neutral and charged particles are produced as result of both hadronic and electromagnetic reactions at the very beginning of the process. These spallation products travel along the beam line interacting with its components producing new radiation and particles disturbing the neutron beam produced. At n_TOF experiment, the contribution of these particles and radiation to the detectors placed in the experimental areas is an inconvenient, since it blinds the detectors used for an specific lapse of time that depends, among others, on their nature and their associated electronics, preventing the measure of the most energetic neutrons released during the spallation process.See Fig. 6.8 Figure 4.23.: g-flash signals produced 118 4.4. γ-flash study Regarding the γ-production, two components are differentiated: prompt, before TOF < 1 µs , mainly formed by γ-rays produced during the spallation process and delayed, after TOF > 1 µs, formed by γ-rays from radiative neutron capture reactions occurring in both the moderator and the cooling system as well as from material surrounding the spallation target. The γ-flash refers the prompt radiation background explained above as well as charged particles plus a halo of both charged particles and radiation coming from the interaction of the high-energy neutrons and γ-rays with the second collimator showed in the beam line. For EAR1: Reduction of the section of the beam pipe passing through the concrete wall (performed in 2014). Taking into account considerations to reduced the background, the socalled γ-flash is observed in both EARs. The γ-flash is taking as a time stamp reference for the ToF since it is the first signal detected by the acquisition frame for each proton bunch impinging on the spallation target, being the signal provided by the PKUP the trigger to start the acquisition system for each channel. In this work, the detectors CeBr3-PMT, CsI(Na)-PMT and CeF3_L50- PMT were used to characterised the γ-flash using sensitive detectors to γ radiation and charged particles, whose reduced active volumes make them appropriated for counting and timing measurements in this environment with such a so high background but not for energy deposition because its low peak to Compton ratio, as it was explained in Chapter 2 4.4.2. γ-flash measurements CeBr3-PMT and CsI(Na)-PMT were placed to measure the g-flash and find out its characteristics when it arrives to an inorganic scintillator crystal. The γ-flash is the first signal, in timing terms, as it states above, that is registered by the detectors. The γ-flash is taking as a time stamp reference for the ToF since it is the first arriving signal at the speed of light. When detected it can be used to accuratelly know the energy of the neutrons 119 4. Beam-tests in the n_TOF experiment arriving inside the acquisition frame after every proton bunch. These signals are registered by an digitiser module that samples the signals. In this case, the sampling rate for both detectors is 1/2ns, i.e. frecuency. The γ-flash signal blinds the detectors at the very beginning of the measurement. The lapse of time of the γ-flash depends on the nature of the detector and their associated electronics. Both the CeBr3-PMT and CsI(Na)-PMT shows signal outputs both electrons and gammas and identical signal shapes, that means, that for the initial hypothesis for γ radiation are appropriate to analysis the structure of the γ-flash. During 2012 n_ ToF campaign, a devoted measurement to study the gamma-flash was performed using mainly scintillator detectors. In this chapter, the results and measurements made/performed during that campaign are explained/exposed and showed. During the 2012 campaign CeBr3-PMT was placed parallel to the concrete wall that separates the Secondary Area to the EAR1 on the right side, following the direction of the neutron beam. The detectors, were placed 20 cm from the centre of the beam. The neutron beam travels along the experimental area through air. The collimator uses was the capture mode: 18 mm diameter. Regarding the signals/waveforms acquired by such scintillator and comparing them with the acquired waveforms that coming from a γ-ray source, it is obvious that the set of γ-flash signals shown in Fig. X, are wider in terms of timing width as well as higher in terms of amplitude. Besides, it sometimes observe that the γ-flash can present a different waveform structure from bunch to bunch of protons; that observation leads to supposed that the γ-flash is formed by several individual γrays and charged particles, mainly electrons, that reach/achieve the EAR1 faster than the scintillators are able to resolved in terms of timing, even using a fast system detection like the CeBr3-PMT. First, a detailed observation of the original pulses, see Fig X shows that: •Pile-up is present in some pulses •The width is wider than the γ-ray registered signal a γ-ray source in the oscilloscope. 120 4.4. γ-flash study Figure 4.24.: CeBr3-PMT and CsI(Na)-PMT detectors placed at EAR1. 121 4. Beam-tests in the n_TOF experiment (a) (b) (c) (d) Figure 4.25.: γ-flash pulses from CeBr3-PMT during the 2012 detector test for γ-flash. 122 4.4. γ-flash study Figure 4.26.: Amplitude distribution of the γ-flash for the first and second pulses. •The signals shows a different structure/shape from bunch to bunch of protons. •A fraction of the pulses are larger than the capability of the acq system. •Signals after the first registered are also produced in some acquisitions. Therefore, to analyse the γrays a set of observables have been defined based on the observations of the CeBr3-PMT produced. The observables which will provide information about are related to the amplitude and time of the signals produced. In this case, the pulses are well-defined and no smooth is needed to improved the signal contour, thus, the definition of the observables is accurate through the analysis of the raw waveform. The observables chosen are the initial time of the signal, the maximum amplitude both the first and second pulses registered respectively. 123 4. Beam-tests in the n_TOF experiment (a) (b) (c) Figure 4.27.: γ-flash profiles for differen deposited energy ranges (related to amplitude distribution in Fig. 4.26) (a) Single pulses profiles. (b) Two or more pulses profiles. (c) γ-ray profile. Note the width of the signal produced by the γ-flash single contribution is larger than the a gamma-ray pulse. 124 4.4. γ-flash study Figure 4.28.: Timing distribution of the γ-flash for the first and second pulses. Three component are distinguished: very prompt, prompt and delayed. 1. The amplitude distribution of the first-arriving pulses shows an accumulation ranging from 0.2 to 5 volts. A 10 % of the firsts signals are saturated. See Fig 4.26 2. Similarly to the second pulses distribution. 3. Concerning timing, three components are distinguished: very prompt, from 0 to 10ns, prompt, from 10 to 50 ns and delayed, >50 ns. See 4.28 Experiments dedicated to fission using PPACs detectors were performed in 2012. In parallel, both CeBr3-PMT and CsI(Na)-PMT were placed at 40 cm far from centre of the beam. As the second collimator system was modified to have a diameter of 80 mm, larger than the one used during the capture detectors test, the γ-flash signals were all saturated in both scintillators. See Figs. 4.29 In the CsI(Na)-PMT case, a rebound is observed after the decay time of the pulses because a fADC problem arising when the input signals exceeds the maximum amplitude allowed. Furthermore, a dead time of the order of 10 µs appears, what is not observed in the no-saturated pulses, Fig. 4.30. In order to overcome this problem, a signal attenuator has been designed. A more detailed description of the attenuator is given in the Apendix D. 125 4. Beam-tests in the n_TOF experiment (a) (b) Figure 4.29.: γ-flash pulses from CeBr3-PMT (a) and CsI(Na)-PMT (b) during the 2012 PPAC fission measurements for γ-flash. The effectiveness of such attenuator can be observed in Fig. 4.31, where the attenuated signals appears without the rebound and saturation problems. During the 2014 commissioning campaign new measurements were performed in both EARs. Besides the CeBr3-PMT, a very fast plastic scintillator was placed in parallel with the CeBr3-PMT in order to measure the γ-flash in coincidence. The advantages of these set-up is, since the plastic scintillator is very thin and as a consequence its detection efficiency for γrays is negligible but not for charged particles, that the analysis the data of both detectors provides a qualitative measure and separate of the contribution of the γrays and charged particles. Besides these detectors, a third one is added to the measured system: a CeF3_L50-PMT detector,described in Chapter 2. This new set-up measured the γ-flash in EAR2 during the 2014 commissioning. See Fig.4.32 126 4.4. γ-flash study Figure 4.30.: γ-flash pulses from CsI(Na)-PMT. Note the rebound after the decay time of the saturated signals. Figure 4.31.: γ-flash attenuated pulses from CsI(Na)-PMT using a Zener diode based attenuator. Note that the rebound after the decay time of signals is removed. 127