UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A low gain fine mesh photomultiplier tube for pure CsI Fujimoto, Derek Jun 2015

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


24-ubc_2015_november_fujimoto_derek.pdf [ 47.08MB ]
JSON: 24-1.0167726.json
JSON-LD: 24-1.0167726-ld.json
RDF/XML (Pretty): 24-1.0167726-rdf.xml
RDF/JSON: 24-1.0167726-rdf.json
Turtle: 24-1.0167726-turtle.txt
N-Triples: 24-1.0167726-rdf-ntriples.txt
Original Record: 24-1.0167726-source.json
Full Text

Full Text

A Low Gain Fine MeshPhotomultiplier Tube for Pure CsIbyDerek Jun FujimotoB.Sc., McGill University, 2013A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinThe Faculty of Graduate and Postdoctoral Studies(Physics)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)October 2015c© Derek Jun Fujimoto 2015AbstractThe increased luminosity of the upgraded SuperKEKB accelerator in turnmandates an upgrade to the Belle detector. One proposed upgrade is toexchange the existing thallium doped cesium iodide scintillation crystals(CsI(Tl)) in the endcap calorimeter with pure cesium iodide (CsI). Oneadvantage of pure CsI is its shorter decay time constant. This would reducethe amount of time taken to process each event, which in turn reduces thechance of simultaneously measuring the energy of different two particles(pileup). Hamamatsu Photonics has produced the R11283 photomultipliertube with a nominal average gain of 255± 11, ideal for measuring the lightproduced by scintillation in pure CsI while in a magnetic field. A prototypearray of 16 photomultiplier tubes was built and tested at TRIUMF. Thiswork documents the characterization of the photomultiplier tube as wellas University of Montreal’s pre-amplification and shaper electronics. Theprimary results can be split into four distinct measurements: the electronicnoise, the short term stability, the excess noise factor, and the lifetime. Theelectronic noise was initially measured with cosmic rays and was found to be(77± 2) keV using a Belle II pure CsI crystal. The short term stability wasmeasured with a set of calibration sources, and the variation over a weekwas (0.28 ± 0.03)% after temperature corrections. The excess noise factorwas found to be (1.9 ± 0.1 ± 0.4) using a pulsed UV laser. This result wasaccompanied by an additional electronic noise measurement of 1730 ± 33electrons at the anode. The lifetime was found using a UV LED array anda 207Bi source, with the gain × quantum efficiency reduced to (93 ± 3)%after about 48 days of aging in real time. This was equivalent to 70 yearsof standard Belle II operation with 7 C having passed through the anode.There were several sets of aging behaviours observed, with some evidencethat the anode charge is not the sole factor in aging.iiPrefaceWhile the motivation for the work presented is due to the Belle II Col-laboration, my supervisor Christopher Hearty was responsible for most ofthe direction of the project, for dictating which measurements needed to betaken, and for providing a blueprint on how each measurement should beaccomplished. He also provided support in the implementation of the setup.I was responsible for the implementation of his ideas, both in the setup andthe analysis unless otherwise indicated. A lot of the work done in chapter 3was prefaced by Chris and Eddie Ji [1][2].Chelsea Dunning procured and light-proofed the dark box that was usedin all measurements, as well as the incubator. She also helped with some ofthe setup in chapter 4, where she determined the best desiccant, helped pre-pare the incubator, and provided instructions for mixing the silicon rubberused in chapter 6. Pierre Amaudruz of TRIUMF put together the initial MI-DAS readout and helped debug issues with the CAMAC crate. He was alsoresponsible for building the pulsed UV laser, and provided the optics usedin chapter 6. Jean-Pierre Martin and Nikolas Starinski of the Universite´de Montre´al designed and built both versions of the preamp, the shapers,motherboards, and a high voltage divider. Version 4 of the preamp, and theshapers were only used in chapter 7, and the V3 preamp was used in thepreceding chapters. The mixing of the silicone rubber and the gluing of theCsI pucks to the PPs was done with the aid of Dylan Jow. Many of the de-tector facility staff helped with the procurement of materials, as well as withsuggestions for setup design. Notable persons include Wayne Faszer, PhilipLu, Peter Vincent, and Robert Openshaw. I designed and constructed allof the specialized setup pieces: the support structures for the excess noisemeasurement, the array support structure for the aging measurement, andmodifications to the incubator and the dark box. With Chris’ direction, Idesigned and implemented all of the circuits used in the presented work,with the exception of that in chapter 3.All of the analysis was performed in C++ using the ROOT frame-work, developed at CERN. Many of the modifications to the MIDAS andROOTANA readout code were also implemented by me, in particular foriiiPrefacethe measurements after, and including, chapter 6.ivTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . vList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ixList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Belle Physics . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 The Belle II Endcap Calorimeter . . . . . . . . . . . . . . . . 22 The Low Gain Fine Mesh Photopentode . . . . . . . . . . . 52.1 On the Operation of Photomultiplier Tubes . . . . . . . . . . 52.2 The R11283 Photopentode . . . . . . . . . . . . . . . . . . . 63 Electronic Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . 83.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 83.2 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . 83.3 Results and Analysis . . . . . . . . . . . . . . . . . . . . . . 103.3.1 The Electronic Noise and the ENE . . . . . . . . . . 103.3.2 The Other Fit Parameters (Signal Waveform Fits) . . 143.3.3 The Other Fit Parameters (Combined Waveform Fits) 163.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164 Short Term Stability . . . . . . . . . . . . . . . . . . . . . . . . 174.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 174.2 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . 174.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20vTable of Contents4.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235 Hamamatsu Measurements . . . . . . . . . . . . . . . . . . . . 245.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 245.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245.2.1 On the Hamamatsu Measured Values . . . . . . . . . 245.2.2 Gain vs Operating Voltage . . . . . . . . . . . . . . . 255.2.3 Comparisons . . . . . . . . . . . . . . . . . . . . . . . 275.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306 Excess Noise Factor . . . . . . . . . . . . . . . . . . . . . . . . 326.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 326.2 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . 326.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346.3.1 PMT Single Photon Measurement . . . . . . . . . . . 346.3.2 Time to Stabilization . . . . . . . . . . . . . . . . . . 356.3.3 Uncertainty of Measurement . . . . . . . . . . . . . . 376.3.4 Calibration Pulse Measurement . . . . . . . . . . . . 376.3.5 Excess Noise Factor . . . . . . . . . . . . . . . . . . . 386.3.6 Excess Noise at Reduced Operating Voltage . . . . . 406.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417 Aging and Lifetime . . . . . . . . . . . . . . . . . . . . . . . . . 427.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 427.2 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . 437.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 447.3.1 Current Baseline and Estimation . . . . . . . . . . . . 447.3.2 Aging and Lifetime . . . . . . . . . . . . . . . . . . . 447.3.3 Post-Aging Stability . . . . . . . . . . . . . . . . . . . 527.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 568 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60AppendicesA Circuit Diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . 65A.1 Electronic Noise . . . . . . . . . . . . . . . . . . . . . . . . . 65A.2 Short Term Stability . . . . . . . . . . . . . . . . . . . . . . 66viTable of ContentsA.3 Excess Noise Factor . . . . . . . . . . . . . . . . . . . . . . . 67A.4 Aging and Lifetime . . . . . . . . . . . . . . . . . . . . . . . 69B Equipment Specifications . . . . . . . . . . . . . . . . . . . . . 70B.1 Electronics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70B.1.1 U. de Montre´al Preamp V3 . . . . . . . . . . . . . . . 70B.1.2 U. de Montre´al Preamp V4 and Motherboards . . . . 71B.1.3 Modifed Preamp . . . . . . . . . . . . . . . . . . . . . 73B.1.4 LeCroy L2249 and L2259B . . . . . . . . . . . . . . . 74B.2 Optics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75B.2.1 Optical Grease . . . . . . . . . . . . . . . . . . . . . . 75B.2.2 Silicone Rubber . . . . . . . . . . . . . . . . . . . . . 75B.2.3 Standard High-Precision PMT . . . . . . . . . . . . . 76B.2.4 Pulsed UV Laser . . . . . . . . . . . . . . . . . . . . . 77B.2.5 Diffuse Reflective Screen . . . . . . . . . . . . . . . . 78B.2.6 McGill LED Array . . . . . . . . . . . . . . . . . . . 79B.3 Miscellaneous Equipment . . . . . . . . . . . . . . . . . . . . 79B.3.1 Dark Box . . . . . . . . . . . . . . . . . . . . . . . . . 79B.3.2 Desiccant . . . . . . . . . . . . . . . . . . . . . . . . . 80B.3.3 BINP CsI Crystal . . . . . . . . . . . . . . . . . . . . 81B.3.4 Temperature and Humidity Probe . . . . . . . . . . . 81B.3.5 Incubator . . . . . . . . . . . . . . . . . . . . . . . . . 81B.3.6 Calibration Sources . . . . . . . . . . . . . . . . . . . 83C Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84C.1 Novosibirsk . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84C.2 Excess Noise Factor . . . . . . . . . . . . . . . . . . . . . . . 85C.3 Excess Noise Factor: Why the PMT Was Not Useful . . . . 85C.3.1 Method 1: The Direct Measurement of Npe . . . . . . 86C.3.2 Method 2: An Alternate Expression for Npe . . . . . 87C.3.3 Conclusion: PMT Not Useful . . . . . . . . . . . . . . 87C.4 Ratio of the Quantum Efficiencies . . . . . . . . . . . . . . . 88D Additional Results . . . . . . . . . . . . . . . . . . . . . . . . . 89D.1 Electronic Noise: Effect of Counter Timing . . . . . . . . . . 89D.2 Residuals for Energy Linearity . . . . . . . . . . . . . . . . . 90D.3 Gain as a Function of Operating Voltage . . . . . . . . . . . 90D.4 Calibration Pulse Results in Alternate Units . . . . . . . . . 91D.5 Justification of Novosibirsk Use in Excess Noise Factor RawFits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92viiTable of ContentsD.6 Screenshot of Calibration Pulse Signal . . . . . . . . . . . . . 93D.7 Details on the Aging Measurement . . . . . . . . . . . . . . . 94viiiList of Tables5.1 Hamamatsu Measured Quantities . . . . . . . . . . . . . . . . 31B.1 Calibration Source Decay Energies . . . . . . . . . . . . . . . 83D.1 Photopentode Categorization Criteria . . . . . . . . . . . . . 95ixList of Figures1.1 Drawing: Belle II Wireframe . . . . . . . . . . . . . . . . . . 32.1 Drawing: PMT Electron Multiplication . . . . . . . . . . . . 52.2 Drawing: Fine Mesh Dynode . . . . . . . . . . . . . . . . . . 62.3 Photo: Photopentode and U. Montre´al Preamp V3 . . . . . . 73.1 Drawing: Electronic Noise Experimental Setup . . . . . . . . 93.2 Photo: Electronic Noise Setup . . . . . . . . . . . . . . . . . . 93.3 Screenshot: Electronic Noise Signal Waveform . . . . . . . . . 103.4 Plot: Signal Waveform Maximum Value Distribution . . . . . 113.5 Plot: Average Signal Waveform . . . . . . . . . . . . . . . . . 123.6 Plot: PDF Fit to Typical Signal Event . . . . . . . . . . . . . 123.7 Screenshot: Electronic Noise Empty Waveform . . . . . . . . 123.8 Plot: Distribution of Heights From Fit to Noise + Average . 133.9 Plot: Distribution of Heights From Fit to Signal . . . . . . . 143.10 Plot: Distribution of Pedestals From Fit to Signal . . . . . . 153.11 Plot: Distribution of Time Offsets From Fit to Signal . . . . 153.12 Plot: Time Offset Check . . . . . . . . . . . . . . . . . . . . . 153.13 Plot: Voltage Offset Correlation . . . . . . . . . . . . . . . . . 164.1 Drawing: Stability Setup Diagram . . . . . . . . . . . . . . . 174.2 Picture: CsI and Photopentode Setup for Stability . . . . . . 184.3 Screenshot: Stability Photopentode Waveform with Gate . . 194.4 Plot: 207Bi Spectrum . . . . . . . . . . . . . . . . . . . . . . . 194.5 Plot: 137Cs Peak with Novosibirsk + Exponential Fit . . . . . 204.6 Plot: Energy to ADC Calibration . . . . . . . . . . . . . . . . 214.7 Plot: Short Term Stability with Temperature Overlay . . . . 224.8 Plot: Peak Response as a Function of Temperature . . . . . . 224.9 Plot: Short Term Peak Response Corrected For Temperature 224.10 Plot: Projection of Temperature Corrected Peak Responses . 225.1 Plot: Cathode and Anode Luminous Sensitivities . . . . . . . 25xList of Figures5.2 Plot: Cathode Blue Sensitivity Index . . . . . . . . . . . . . . 255.3 Plot: Peak Location vs Operating Voltage . . . . . . . . . . . 265.4 Plot: Histogram of Gains at −1000 V . . . . . . . . . . . . . 275.5 Plot: Average Gain as a Function of Operating Voltage . . . 275.6 Plot: Anode and Cathode Sensitivity Correlation . . . . . . . 275.7 Plot: Peak vs Energy Slope and Anode Sens. Correlation . . 275.8 Plot: Peak vs Energy Slope and Cathode Sens. Correlation . 285.9 Plot: Peak vs Energy Slope and Blue Index Correlation . . . 285.10 Plot: Peak vs Energy Slope and Gain Correlation . . . . . . . 295.11 Plot: Anode Sens. vs Gain at −750 V . . . . . . . . . . . . . 295.12 Plot: Cathode Luminous Sensitivity as a Function of Gain . . 295.13 Plot: Peak vs Energy Slope/Gain and Cat. Sens. Correlation 295.14 Plot: Average Resolution vs Energy . . . . . . . . . . . . . . 306.1 Photo: Excess Noise Setup Top View . . . . . . . . . . . . . . 336.2 Photo: Excess Noise Setup Front View . . . . . . . . . . . . . 346.3 Plot: PMT Single Photon Measurement . . . . . . . . . . . . 356.4 Plot: Raw Spectrum From Laser . . . . . . . . . . . . . . . . 366.5 Plot: Time to Stabilization . . . . . . . . . . . . . . . . . . . 376.6 Plot: Stability of Pulses from UV Laser . . . . . . . . . . . . 376.7 Plot: Uncertainty in Measurement . . . . . . . . . . . . . . . 376.8 Plot: Calibration Pulse ADC vs C . . . . . . . . . . . . . . . 386.9 Plot: Excess Noise Factor . . . . . . . . . . . . . . . . . . . . 396.10 Plot: Excess Noise Factor Histogram . . . . . . . . . . . . . . 396.11 Plot: Electronic Noise Histogram . . . . . . . . . . . . . . . . 396.12 Plot: Equivalent Noise Energy Histogram . . . . . . . . . . . 406.13 Plot: Excess Noise vs Operating Voltage . . . . . . . . . . . . 407.1 Photo: Aging Photopentode Array with CsI . . . . . . . . . . 427.2 Photo: Aging Setup . . . . . . . . . . . . . . . . . . . . . . . 437.3 Plot: Integrated Current During Aging . . . . . . . . . . . . . 447.4 Plot: Aging LED Pulse Spectrum . . . . . . . . . . . . . . . . 457.5 Plot: 207Bi Spectrum . . . . . . . . . . . . . . . . . . . . . . . 457.6 Plot: 207Bi Peak Locations as a Function of Energy . . . . . . 457.7 Plot: Temperature and Humidity Readings Throughout theAging Measurement . . . . . . . . . . . . . . . . . . . . . . . 467.8 Plot: Peak Temperature Dependence . . . . . . . . . . . . . . 477.9 Plot: Control PP and Aged PP Relative Change in Gain . . . 477.10 Plot: Relative Performance of PP - Aging Only (1/2) . . . . 497.11 Plot: Relative Performance of PP - Aging Only (2/2) . . . . 50xiList of Figures7.12 Plot: Relative Post-Aging Gain of All Photopentodes . . . . . 517.13 Plot: Relative Performance of PP - Includes Post-Aging (1/2) 537.14 Plot: Relative Performance of PP - Includes Post-Aging (2/2) 547.15 Plot: Post-Aging Stability Histogram . . . . . . . . . . . . . . 557.16 Plot: Post-Aging RMS/Mean of All Photopentodes . . . . . . 55A.1 Circuit: Electronic Noise . . . . . . . . . . . . . . . . . . . . . 65A.2 Circuit: Short Term Stability . . . . . . . . . . . . . . . . . . 66A.3 Circuit: Calibration Pulse . . . . . . . . . . . . . . . . . . . . 67A.4 Circuit: Excess Noise Factor . . . . . . . . . . . . . . . . . . . 68A.5 Circuit: Aging and Lifetime . . . . . . . . . . . . . . . . . . . 69B.1 Photo: U. Montre´al Voltage Divider . . . . . . . . . . . . . . 70B.2 Photo: U. Montre´al Preamp V4 . . . . . . . . . . . . . . . . . 71B.3 Photo: U. Montre´al Motherboard . . . . . . . . . . . . . . . . 72B.4 Photo: U. Montre´al Modifed Preamp . . . . . . . . . . . . . . 73B.5 Plot: Current from Modified Preamp as a Function of HV . . 74B.6 Plot: ADC Linearity Test . . . . . . . . . . . . . . . . . . . . 75B.7 Photo: Gluing CsI to Photopentode . . . . . . . . . . . . . . 76B.8 Photo: Diffuse Reflective Screen . . . . . . . . . . . . . . . . 78B.9 Photo: McGill UV LED Array . . . . . . . . . . . . . . . . . 79B.10 Photo: Dark Box . . . . . . . . . . . . . . . . . . . . . . . . . 80B.11 Photo: Temperature and Humidity Probe . . . . . . . . . . . 81B.12 Photo: Incubator . . . . . . . . . . . . . . . . . . . . . . . . . 82C.1 Plot: Npe (PP) vs Npe (PMT) . . . . . . . . . . . . . . . . . 88D.1 Plot: Multi-peak Time Offsets to PDF Fits in Chapter 3 . . . 89D.2 Plot: Energy Linearity Residuals . . . . . . . . . . . . . . . . 90D.3 Plot: Gain vs Operating Voltage for all Photopentodes . . . . 90D.4 Plot: Histogram of the Slopes of Gain vs Operating Voltage . 90D.5 Plot: Calibration Pulse Alt. Units: ADC v e− . . . . . . . . . 91D.6 Plot: Calibration Pulse Alt. Units: mV v C . . . . . . . . . . 91D.7 Plot: Calibration Pulse Alt. Units: mV v e− . . . . . . . . . 92D.8 Plot: Fits to Laser Pulse Spectra - Comparison . . . . . . . . 92D.9 Plot: Residuals to Fits for Laser Pulse Spectra - Comparison 93D.10 Screenshot: Calibration Pulse . . . . . . . . . . . . . . . . . . 93D.11 Plot: Current in Modified Preamp as a Function of Time . . 94D.12 Plot: Three Peak 207Bi Fit . . . . . . . . . . . . . . . . . . . 94D.13 Plot: Relative Photopentode Performance with Categoriza-tion Characteristics . . . . . . . . . . . . . . . . . . . . . . . . 95xiiAcknowledgementsI would like to extend my thanks to my supervisor, Christopher Hearty, forhis guidance and instruction. Additionally, in recognition of their supportand advice: my fellow students and friends, with particular acknowledge-ment to Alon Hershenhorn for idea brainstorming and enforced coffee/teabreaks, as well as Jean-Franc¸ois Caron for his useful insights into ROOT andC++. The TRIUMF support staff have also earned my gratitude for theirhelp in DAQ and setup, in particular: Wayne Faszer, Philip Lu, Peter Vin-cent, Robert Openshaw, and Pierre Amaudruz. Lastly, for their continuedsupport, my parents.xiiiChapter 1Introduction1.1 Belle PhysicsThe concept of the atom has existed since the time of ancient Greece, butonly recently has the field of subatomic particle physics emerged. One ofthe first in this field was Rutherford’s gold foil experiment [3]. He showedthat the atom was composed of at least two separate entities: the nucleusand the electron. Since then, particle physicists have worked to find thecomplete set of fundamental particles and to refine the laws that governtheir interactions: the standard model.In their quest for new physics, modern particle physics experiments havetwo primary strategies: direct searches and indirect searches. While bothmake use of particle colliders, direct searches for new physics focus on theproduction of new forms of matter by increasing the collision energy. Incontrast, indirect searches focus on the observation of extremely rare pro-cesses where deviations from the expected proportions of collision productscan be traceable to new physics in the underlying mechanism. The indirectsearches for new physics are often termed flavour physics, as they are thecareful study of the interactions and combinations of each species, or flavour,of particle.The KEK B-factory is an asymmetric e+e− collider situated in Tsukuba,Japan [4], with the goal of producing a large number of B mesons for thestudy of flavour physics. The detector at the KEKB interaction point, Belle,was designed in particular to observe CP violation in B mesons [5], andfrom 1999 to 2009 the Belle detector acquired an integrated luminosity of1000 fb−1 [6]. While most of this data set was at the Υ(4S) resonance, theideal location to produce BB¯ pairs vital to the study of B mesons, Belle wasalso able to provide a series of unique data sets on other Υ resonances [5].In addition to increasing the precision of many standard model parameters,Belle, together with its sister experiment BABAR [7] at SLAC, providedevidence of CP violation in 2001 [8][9] resulting in the 2008 Nobel Prize forphysics and the theory of CP violation [10].The next generation of high luminosity machines will again push the11.2. The Belle II Endcap Calorimeterboundaries of known physics. The next iteration of the KEKB acceler-ator, SuperKEKB, will be well positioned to study energy scales beyondthose attainable by current generation direct search programs. For exam-ple, flavour-changing neutral-current processes are known to be particularlysensitive to new physics which does not have the same suppression mecha-nisms as is present the standard model [11]. A high luminosity experimentstudying phenomenon of this type will therefore be sensitive to the samemechanisms, even if these mechanisms deviate from the standard model atextremely large energy scales.The SuperKEKB asymmetric e+e− collider has a luminosity goal of8 × 1035cm−2s−1 [6], a factor of 40 increase over the luminosity achievedby KEKB. SuperKEKB uses the same beam pipe as KEKB (other than atthe interaction point) and the corresponding upgrade to Belle, Belle II, willmake use of the existing Belle structure. The increased luminosity presentsmany challenges for Belle II. Relevant to this body of work are the pileup in,and the irradiation of the endcap calorimeter (endcap ECL). It is estimatedthat the background rates in the ECL will be a factor of twenty larger withthe increased luminosity [12]. Despite the continued performance of theBelle scintillation crystals [13], it was proposed by the collaboration thatthe thallium-doped cesium iodide (CsI(Tl)) scintillation crystals should bereplaced with pure CsI [6]. Pileup occurs when two or more particles inter-act with the crystal within the crystal’s decay time. In this case the energyof the two particles is additive, resulting in a much larger signal than wouldotherwise result. Pure CsI has a much shorter scintillation time constantthan CsI(Tl) [14][15][16] which will help reduce the number of pileup events,important for missing energy studies [6].1.2 The Belle II Endcap CalorimeterAs seen in figure 1.1, the endcap calorimeter is located at either end of thebarrel section. The main tasks assigned to the endcap ECL are: the detec-tion and determination of photon energy and angular coordinates, electronidentification, trigger generation, luminosity measurement, and to assist inK0L detection [6].The Belle endcap ECL contained 8736 CsI(Tl) crystals, weighing a totalof 43 tons, each a truncated pyramid in a variety of proportions to cover91% of the total 4pi solid angle [18]. Each crystal has a cross section ofabout 5.5 cm × 5.5 cm and a length of 30 cm. Each was wrapped in aGore-Tex film layer and an aluminized mylar layer [13]. The optical readout21.2. The Belle II Endcap CalorimeterFigure 1.1: Belle II detector with endcap ECL highlighted [17].was performed by two PIN photodiodes, each with its own preamp directlymounted to the crystal [18]. Over the course of four years of operation,the cumulative dose in the endcap scintillation crystals was about 100 rad.With the radiation levels expected increase by a factor of 20 for SuperKEKB,it is expected that the light output of the CsI(Tl) will decrease by about25% [18]. Although this is not a pressing concern, the pile up noise due tothe increased particle rate and the slow scintillation time of CsI(Tl) pose amore serious problem. Therefore, major improvements could be made to theendcap ECl by replacing the existing CsI(Tl) in the endcaps with a fasterversion of the crystal.The Belle II Technical Report details three options for the upgrade tothe endcap calorimetry [6]:• Pure CsI crystals with a photomultiplier tube readout.• A lead tungstate crystal, dubbed PWO-II by the PANDA experiment[19], with an avalanche photodiode (APD) readout.• Bi4Si3O12 crystals with APD readout.The most promising option at this time is the pure CsI, due to its mod-erate cost, good performance, and ease of production. The proposal for the31.2. The Belle II Endcap Calorimeterpure CsI option was to keep the existing Belle support structure and bar-rel CsI(Tl) crystals, but to replace the endcap crystals with pure CsI andupgrade all readout electronics [20].A large part of the Canadian contribution has been to the research of thepure CsI option. CsI has two components to its scintillation time constant: afast component of about 30 ns [16], and a slow component of about 1µs [14].This is in comparison to CsI(Tl) which has a one component time constantof 1µs [15]. It should be noted, however, that the light output of pure CsIis about one tenth of CsI(Tl) [6]. To keep the equivalent noise energy atthe level obtained during the Belle experiment, the PIN photodiodes willhave to be replaced with the aforementioned photomultiplier tube readout,as the noise to signal ratio is lower. Additionally, the PIN diodes are notsensitive to UV, making the photopentodes are better suited to read out thespectrum of pure CsI.The work presented is an examination of a proposed photomultipliertube for use with the pure CsI option. All of the measurements were carriedout in the detector facility at TRIUMF and share some basic features: thephotomultiplier tube was kept in a dark box, which was grounded to reducepickup and noise levels. Also, the temperature and humidity of the boxwas continually recorded, and the humidity was controlled with a desiccant.This was to reduce damage to the surface of the slightly hydroscopic CsI.4Chapter 2The Low Gain Fine MeshPhotopentode2.1 On the Operation of Photomultiplier TubesA photomultiplier tube (PMT) is a device for detecting low levels of light,and in some cases, single photons. As seen in figure 2.1, a PMT has threeprimary components: the photocathode, the anode, and the dynodes.PHOTOCATHODEFACEPLATEDIRECTIONOF LIGHTe-ELECTRON MULTIPLIER(DYNODES)FOCUSING ELECTRODELAST DYNODE STEM PINVACUUM(~10P-4)SECONDARYELECTRONANODESTEMFigure 2.1: The electron multiplication process in a PMT [21].The photocathode converts a single photon into a single electron withsome probability (the quantum efficiency). The photocathode is a semicon-ductor and can be described using a band model. Electrons in the valenceband of the photocathode can absorb energy from an incident electron andbecome excited. With enough photon energy, the election is emitted fromthe material and is free to propagate. Electrons emitted from the photo-cathode are called the photoelectrons and are guided towards the dynodesby means of an electric field and careful dynode placement [21].The dynodes are the portion of the PMT that multiply the electrons52.2. The R11283 Photopentodeinto a signal that is readily measurable. For every electron incident on eachdynode, an average of δ > 1 secondary electrons are ejected. The followingcascade of electrons results in the multiplication of the original signal. Thedynode collection efficiency is determined by the material and the geometryof the dynode, while the secondary emission ratio (δ) is a function of thevoltage applied to the dynode and the dynode material.The anode of the PMT is the final stage of the electron multiplicationprocess. It collects the electrons after the dynode cascades and outputs theelectron current, which is measured. In this way, the charge at the anode isproportional to the amount of light incident on the PMT.The gain of the PMT is the factor by which the number of photoelectronshas been multiplied at the anode. High sensitivity PMTs can have 19 dynodestages with gains on the order of 106.2.2 The R11283 PhotopentodeFine meshElectronElectronFigure 2.2: Thefine mesh dynode[21].Hamamatsu Photonics has produced a PMT thatclosely matches the specifications outlined in theBelle II Technical Report [6]. This PMT has five flyingleads to power the dynodes and to read out the currentfrom the anode. Because of the five leads, this PMThas been given the nickname “photopentode” (PP) bythe collaboration. The PP (model R11283) has threefine mesh dynodes (fig. 2.2) which allows it to havea high immunity to magnetic fields. This is becausethe dynodes can be placed in close proximity to eachother, reducing the path length between dynodes. Un-like more typical PMTs whose dynodes are solid plates,the fine mesh has a poor electron collection efficiencyand electrons can bypass a dynode entirely, contribut-ing to the noise of the signal. The factor by whichthis process increases the square of the resolution iscalled the excess noise factor (F ). The measurement ofthe excess noise factor is described in chapter 6. TheR11283 PP has a UV transparent, 2 inch diameter win-dow, closely matching the size of the CsI crystals in theBelle ECL. In particular, the PP is responsive to light with wavelengths from185 nm to 650 nm, with maximal response at 420 nm [22]. This makes itideal for recording the output of pure CsI, as the maximum of the emitted62.2. The R11283 Photopentode(a) Window (photocathode) side. (b) Rear (anode) side, with preamp.Figure 2.3: The Hamamatsu R11283 photopentode with version three of theUniversite´ de Montre´al preamp (right).light spectrum of the material is 315 nm [14]. R11283 is a head-on type,with a bialkai photocathode and a typical gain of 255 ± 11 (ch. 5) at themaximum operating voltage of −1000V. The low gain of the PP is to ensurethat the current at the anode does not exceed the voltage divider’s specifi-cations. If the gain is too large, then the dynodes draw too much currentduring the multiplication process and the PP does not perform as expected.The version of the PP used required the assembly of a custom voltagedivider, preamp, and shaper electronics. The voltage divider’s purpose isto take a single input voltage and redistribute it among the dynodes. Thepreamp and the shaper together produce a signal whose amplitude is propor-tional to the amount of charge passing through the anode. These electronicswere designed and produced at the Universite´ de Montre´al [23]. Version 3of the Montre´al preamp can be seen in figure 2.3, attached to the R11283PP (appendix B.1.1). The preamp is situated directly on the flying leads ofthe PP and the preamp signal is sent to the shaper which can be off site,allowing for reduced space requirements in the detector ECL.In addition to the above characteristics, the R11283 is an ideal PMT forthe Belle II ECL because of its relatively low cost when compared to otherPMTs with similar properties.7Chapter 3Electronic Noise3.1 IntroductionThe electronic noise was an important first quantity to measure since itdictated whether or not low energy measurements would be viable. Thissource of random uncertainty is due to the processing electronics, namelythe preamp and the shaper. The primary goal of this measurement wasto find the equivalent noise energy (ENE). This is the electronic noise inunits of energy deposited in the CsI crystal. In this way, the effect of theelectronics on the energy resolution is comparable. The ENE was measuredusing cosmic muons.3.2 Experimental SetupAs with all of the measurements presented in this work, the experimentalsetup was encased in a light-tight box (appendix B.3.1) which provided a lowlight environment for the phototubes, restricted air flow so that the interiorcould be kept at a low humidity, and shielded the setup from pickup. Amolecular sieve desiccant (appendix B.3.2) was used to keep the humidityto about 10% relative humidity within the dark box.Two plastic scintillators with standard PMTs were used to generate atrigger which selected for cosmic muons. As seen in figure 3.1, any cosmicparticle which generates a trigger has to travel first through the top plasticscintillator, then through the CsI, the lead bricks, and finally the bottomplastic scintillator. Requiring the coincidence of the two plastic scintilla-tors ensures that the particle was a minimum ionizing muon which travelledthrough the CsI, since the lead prevents all other particles of lower momen-tum from reaching the second scintillator.The CsI crystal used was on loan from the Budker Institute of NuclearPhysics (BINP) and is of the same size and style as a Belle II endcap ECLcrystal (appendix B.3.3). The PP was connected to the crystal via an opticalgrease produced by Saint-Gobain (appendix B.2.1). The setup in the dark83.2. Experimental SetupPPPMTPMTLeadCsI CrystalPlastic ScintillatorPreampPlastic ScintillatorFigure 3.1: Experimental setup (side view). Coincidence on the plastic scin-tillators provide the trigger and lead shielding ensures that the cosmic particleis a minimum ionizing muon.box can be seen in figure 3.2.The PP output from the anode was first processed by the V3 preamp.The signal was then sent to an Ortec 474 Timing Filter Amp which servedas the shaper, and the output waveform was recorded by a Teledyne LeCroyWavePro 740Zi oscilloscope. The Ortec shaper had an integration and dif-ferentiation time constant of 50 ns. Refer to figure A.1 for circuit minutiae.Figure 3.2: Experimental setup. Two plastic scintillators were used on topand on bottom of the CsI crystal for the cosmic muon trigger.93.3. Results and Analysis3.3 Results and Analysis3.3.1 The Electronic Noise and the ENEDue to the limitations of the scope’s dynamic range, the electronic noisewas not measurable directly. At the scales necessary to capture the fullsignal waveform, the uncertainty of each bin was dominated by the scopeuncertainty, whereas the electronic noise could only be resolvable withoutthe muon signal present. The plan was therefore:• Record signal waveforms: a cosmic muon had deposited energy in theCsI.• Find the average signal waveform.• Record empty waveforms: no energy was deposited in the CsI.• Construct a probability distribution function (PDF) from the averagewaveform.• Construct combined waveforms: the sum of the empty waveforms andthe average waveform.• Fit the PDF to the combined waveforms and observe how the emptywaveforms affect the fit parameters.A typical cosmic muon event can be seen in figure 3.3. About 5500 ofthese events were collected over the course of about 15 hours.Figure 3.3: Screenshot of a cosmic muon event (shaper output) in the CsIcrystal from the oscilloscope. Pulse height: ≈2.4 V, pulse length: ≈560 ns.A portion of these were not useful, either because the events were emptyor because the event energy was so large that the oscilloscope range was ex-ceeded. Figure 3.4 shows the distribution of the maximum value of the 5500signal waveforms. It should be noted that the oscilloscope set an artificialoffset to all of the voltage values, to make better use of the oscilloscope’srange. This did not affect the outcome as a pedestal was established byrecording a large portion of the waveform prior to the trigger. Waveformswith maximum values peaking at −5.8 V had no evidence of a cosmic muon,103.3. Results and AnalysisEntries  5536Mean    2.858RMS      1.32Underflow       0Overflow        0Max Value (V)0 2 4 6Number of Entries050100150200250300Figure 3.4: The distribution of the signal waveform maximum values.This was used to discriminate against empty events or PP over-saturation.Events in the range of [-4,0] V survived the cut.whereas waveforms with maximum values near 0.5V had surpassed the rangeof the scope. The values kept were events with maximum values within therange of [−4, 0] V, with 84% of the events surviving the cut.The signal waveforms were then averaged bin-by-bin, with the result seenin figure 3.5. This average waveform eliminated all significant variabilityfrom the signal, including the variability due to the energy deposited in theCsI by cosmic muons. The noise from the electronics was then re-introducedin the combined waveforms.The normalized PDF used in the fits had three parameters: the height (ascaling factor), the pedestal (a voltage offset), and a time offset. A typicalsignal waveform with PDF fit can be seen in figure 3.6. The height of thePDF is proportional to the energy deposited in the CsI crystal.113.3. Results and AnalysisBin Index0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000Voltage (V)6−5.5−5−4.5−4−3.5−3−Figure 3.5: The average signalwaveform after the cut. Thiswas then normalized to havea height of one to make thePDF. Each bin corresponds to0.4 ns.Index0 1000 2000 3000 4000 5000Voltage (V)-6-5.5-5-4.5-4-3.5/ ndf 2χ 4850 / 4897Prob 0.6808height 0.002691±2.687 tOffset 0.1967±-1.958 vOffset 0.0007114±-6.04 Figure 3.6: The PDF generated fromfigure 3.5 fitted to a typical signalevent.Figure 3.7: Screenshot of a empty event (no cosmic event) from the oscil-loscope. Waveforms such as this one were added bin-by-bin to the averagewaveform in figure 3.5.An example of an empty waveform can be seen in figure 3.7. Eventssuch as these were added bin-by-bin to the average waveform from figure3.5. The fits to these sums were then affected by the electronic noise, whilethe variation in muon energy was eliminated. The distribution of the fitheights to the combined waveforms can be seen in figure 3.8, and the widthof this distribution was the electronic noise: (5.89± 0.08) mV.123.3. Results and AnalysisEntries 3001Mean 0.0001104±2.829 RMS 7.807e-05±0.006047 Underflow 0Overflow 1 / ndf 2χ 36.55 / 35Prob 0.3964Constant 5.2±227 Mean 0.000±2.829 Sigma 0.000080±0.005885 Height (V)2.81 2.82 2.83 2.84 2.85Number of Events per Bin of Width 0.001 V050100150200250Figure 3.8: The distribution of heights of the fit of the PDF to the noise +average waveform with Gaussian fit. The standard deviation of the distri-bution was (5.89± 0.08) mV.This value needed to be converted into units of energy. To do so, thesignal waveforms were fitted with the PDF (fig. 3.6) and the peak heightswere histogrammed, as seen in figure 3.9. Since the distribution is boundedby zero, it cannot be accurately described with a Gaussian. The fittedfunction used was an asymmetric Gaussian, dubbed the Novosibirsk functionby the collaboration. The Novosibirsk function is the convolution of a log-normal distribution and the theoretical energy spectrum due to Comptonscattering (appendix C.1).133.3. Results and AnalysisEntries 4657Mean 0.01029±3.245 RMS 0.007274±0.702 Underflow 0Overflow 1 / ndf 2χ 52.04 / 58Prob 0.6953normalization 6.3±182.8 peak 0.014±2.829 width 0.011±0.435 eta 0.0407±-0.6011 Height (V)2 2.5 3 3.5 4 4.5 5 5.5 6Number of Events per Bin of Width 0.042 V020406080100120140160180Figure 3.9: The distribution of the heights from fitting the PDF to the signalevents. This was used to convert the standard deviation in figure 3.8 intounits of energy.Given that the most likely energy deposit due to a muon along the 30 cmlength of the CsI crystal is 190 MeV[24], and that the average crystal heightwas 5.88 cm, the most likely energy deposit in the CsI due to a cosmicmuon travelling along the vertical axis was about 37.2 MeV. This energyvalue corresponds to the peak of the distribution of the signal peak heights(fig. 3.9). The conversion rate from the electronic signal to units of energywas (13.16 ± 0.07) MeV/V. Applying this to the standard deviation of thedistribution of the heights of the combined waveforms (fig. 3.8), the ENEwas found to be (77± 2) keV.3.3.2 The Other Fit Parameters (Signal Waveform Fits)It was also instructive to consider the distributions of the other fit parame-ters. Figures 3.10 and 3.11 respectively show the distribution of the pedestalsand the time offsets from fits to the signal waveforms. In both, there is ev-idence of a bimodal structure. The distribution of the pedestals was likelydue to pickup, and would be a constant factor throughout the entire signalwaveform. This would be corrected for by the pedestal fit parameter.To investigate if the distribution of the time offsets was an artifact ofthe fitting, the leading edge of the signal waveform was fitted with a linearline (fig. 3.12). The distribution of the intersect of the fit and the averagepedestal also produced the bimodal features seen in figure 3.11. Additionally,adding a delay to the signal from either the top or bottom plastic scintillatorsdid not remove this effect. Rather, with the lower scintillator delayed, the143.3. Results and AnalysisVoltage Offset (V)-6.08 -6.07 -6.06 -6.05 -6.04 -6.03 -6.02Number of Events per Bin of Width 0.001 V050100150200250300Figure 3.10: The distribution ofthe pedestals from fitting the PDFto signal events.Time Offset (ns)-10 -5 0 5 10 15 20Number of Events per Bin of Width 0.8 ns050100150200250300Figure 3.11: The distribution ofthe time offset from fitting thePDF to signal events.bimodal distribution from figure 3.11 resulted, whereas if the upper counterwas delayed, the time offset distribution had three, well-separated peaks.This pattern was repeated when a standard PMT was used to read out thecrystal. It was therefore concluded that the bimodal distribution of the timeoffset arose from the timing of the plastic scintillator counters and not theCsI or PP system. See appendix D.1 for a comparison of the time offsetdistributions from this analysis.Vector Index0 1000 2000 3000 4000 5000Voltage (V)-6-5.5-5-4.5-4-3.5-3Figure 3.12: Checking the fitting of the PDF to the signal waveforms. Thebimodal peak structure in figure 3.11 was still present. Each bin represents0.4 ns.153.4. Conclusion3.3.3 The Other Fit Parameters (Combined Waveform Fits)It was seen that the signal height and voltage offset from the fits to thecombined waveforms were correlated. To compare, a set of randomly gen-erated noise data was added to the averaged waveform and was fitted withthe PDF. The lack of a correlation in this case (fig. 3.13), illustrates an issuewith the fitting of the combined waveforms. Since the shaper takes a signalthat approximates a delta function and stretches it into a full waveform, itwas thought that the bins of the measured noise waveforms would containoverlap from neighbouring bins and would not be fully independent. Thiswould be in violation of one of the base assumptions in the minimizationpackage provided by ROOT. To do this properly, the chi-squared would haveto be calculated and minimized using the full covariance matrix.-0.006 -0.004 -0.002 0 0.002 0.004 0.006 0.0082.6452.652.6552.662.6652.672.6752.682.6852.69HeightVoltage Offset (V)(a) Real noise correlation.Voltage Offset (V)-0.02 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02 0.025Height (V)2.6642.66452.6652.66552.6662.66652.667(b) Fake noise correlation.Figure 3.13: Voltage offset correlation comparison for both measured andgenerated noise added to the averaged signal waveform. It was noted thatonly the combined waveform with the measured noise set was correlated.3.4 ConclusionIt was tentatively concluded that the equivalent noise energy of the R11283photopentode, using the Belle II pure CsI crystal from BINP and the Uni-versite´ de Montre´al preamp, was (77± 2) keV for 37.2 MeV cosmic muons.The result was tempered by the correlation found between two of the fit pa-rameters of the fits to the combined noise signals, which suggest that therewas a violation of the assumption that the bins in the noise signal wereindependent. To properly follow this study, the covariance matrix of thewaveforms would have to be constructed and from this the chi-squared canbe calculated and minimized. The ENE was also found in chapter 6 using adifferent methodology and a smaller, high quality CsI crystal.16Chapter 4Short Term Stability4.1 IntroductionThe goal of this measurement was to find the variability in the photopentodeand electronics output after correcting for variations in temperature. Incontrast to the electronic noise measurement of chapter 3, a light source ofreproducible intensity was needed. This was provided by the scintillation ofgamma rays in a small CsI crystal. The gamma rays were emitted by oneof three calibration sources, and the stability over a week was determinedfrom the signal produced by the CsI crystal. By virtue of using calibrationsources with known decays, the linearity of the PP signal with energy wasalso observed.4.2 Experimental SetupCsIPPShieldSourceIncubatorFigure 4.1: Setup tomeasure the short termstability of the PP,top-down view.As mentioned in the introduction, this measure-ment used a calibration source to provide gammarays to induce scintillation in the CsI crystal. Thecrystal was procured from Saint-Gobain and wasmuch smaller than the BINP Belle II crystal: acylinder of 2.7 cm in diameter and 2.5 cm in length.The small crystal size reduced the number of cos-mic events, as well as to increase the efficiencywith respect to the amount of light collected bythe photopentode (PP). To increase the data collec-tion rate, a peak finding analog to digital converter(ADC - appendix B.1.4) was used in the place of theoscilloscope. The data acquisition (DAQ) softwarewas handled by the MIDAS framework [25]. Theuse of the calibration sources also permitted thecorrelation of the ADC readout to the amount of energy deposited in thecrystal. Figure 4.1 shows a drawing of the setup. An acrylic puck was used174.2. Experimental Setupto shield the crystal from any beta radiation from the calibration source.The whole setup was encased in an incubator (appendix B.3.5) to controlthe temperature as much as possible. The temperature was measured to be(34.5±3) ◦C over the course of the measurement. The incubator was placedwithin the same dark box as was used for the electronic noise measurement.A hole was cut in the corner of the incubator window to let cables passthrough, and any gaps were plugged with foam. Figure 4.2 shows the setupFigure 4.2: Device for holding the PP and CsI crystal in place. The opticalconnection pictured is a rubber silicon cookie, but the measurements used anair gap since it was more reproducible.that was encased in the incubator. While the pictured setup makes use of asilicon cookie for the optical conduit, an air gap was used in the final mea-surement. The air gap was preferred over both the optical grease used inchapter 3 and the cookie because of the reproducibility of the measurement.The cookie had a tendency to detach from the crystal or the glass of thephototube, or to trap air pockets when applied. Both cases left the opti-cal conductivity non-uniform across the surface of the PP. With the opticalgrease, the amount of grease applied affected the output, and the greasereacted with the CsI, yellowing it with time. Of the three, variations in theair gap produced the least variation in output. The air gap was maintainedto less than a millimetre for the experiment, but the size of the gap was notprecisely controlled.184.2. Experimental SetupFigure 4.3: Screenshot of the waveform output of the Ortec 474 shaper, withthe NIM gate used to trigger the ADC. The gate generation was based on adiscriminator threshold set on the shaper output.ADC0 200 400 600 800 1000 1200 1400 1600 1800 2000Number of Entries010002000300040005000Bi207 Raw Spectrum0.57MeV1.064MeV1.77MeVFigure 4.4: 207Bi spectrum as seen by the ADC using the described setup.There are three peaks for 207Bi but one of them has very low statistics. Thecutoff at about bin 75 is the location of the discriminator threshold, and thesmall peak below that is due to a random trigger.Three calibration sources were used to generate signals: 22Na, 137Cs,and 207Bi (appendix B.3.6). The shaper output was amplified to put theoutput peak heights near the center of the range of the peak-sensing ADCused to digitize the signal (appendix B.1.4). After amplification, the signalwas split: one copy was sent to the ADC and the other to a discriminatorused to produce the gate. The full circuit can be found in figure A.2 in theappendix. A signal from the amplifier and the corresponding gate can beseen in figure 4.3.The DAQ software used to process the ADC signals was the MIDAS and194.3. ResultsROOTANA frameworks [25], which were written at TRIUMF (Canada) andPSI (Switzerland). The distributions of the 207Bi peak heights as measuredby the ADC and recorded by MIDAS can be seen in figure 4.4. Since thepeak sensing ADC had a DC offset, the initial plan was to use a randomtrigger and to determine the pedestal from the peak produced by noisereadings. This is the small peak below the discriminator threshold in figure4.4. As seen in appendix B.1.4, the L2259B ADC is non-linear for smallinputs and the pedestals were instead determined from plotting the decayenergy against the ADC peak location.4.3 ResultsDue to backscatter, the spectra from the various calibration sources containa background that was roughly exponential in nature, whereas the peaksthemselves could be described with a Novosibirsk function. Figure 4.5 showsa fit of the sum of a Novosibirsk function and an exponential to the peakof the 137Cs source. The peak location of the Novosibirsk fit was the valueused for the PP response.run00000003Entries 471378Mean 0.1857±454.5 RMS 0.1313±81.42 Underflow 0Overflow 0 / ndf 2χ 325.8 / 344Prob 0.7526normalization 5.6±910.8 peak 0.4±484.3 width 0.43±58.51 eta 0.0051±-0.0441 expScale 0.124±9.443 expPower 0.0004±-0.0103 ADC300 350 400 450 500 550 600 650Number of Entries02004006008001000Figure 4.5: 137Cs spectrum (zoomed in on peak) fitted with the sum of aNovosibirsk function and an exponential. The peak of the Novosibirsk func-tion was used to track the PP response.This was done with all three of the calibration sources to determine thelinearity of the PP with respect to the energy deposited in the CsI. Theanalysis also resulted in the conversion rate from ADC bin to energy, and204.3. Resultsthe pedestal of the ADC for the linear region. The result can be seen infigure 4.6. The intercept of the linear fit was used as the pedestal the thefollowing chapters, and was subtracted from the peak locations. The slopeis the ADC bin to energy conversion rate.Energy Deposited (MeV)0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8ADC Peak Location (bin)Na22Bi207Cs1370200400600800100012001400/ ndf 2χ 99.42 / 4Prob 1.307e-200.847±-78.65 1.251±843.3 SlopeInterceptFigure 4.6: ADC response as a function of energy deposited in the Saint-Gobain pure CsI crystal. The response is linear. See figure D.2 in theappendix for the residuals from the fit.For the short term stability, only the 137Cs source was used. This wasbecause a consistent light source was needed, so only one decay was required,and the lack of additional decay energies meant that the location of the137Cs peak was not affected by backscatter from other decay energies. ThePP peak height spectrum was histogrammed for about a week, with eachhistogram covering one hour. Temperature measurements were made every15 minutes. It is worth noting that the precision of the temperature reader(appendix B.3.4) was only 0.1 ◦C. The peak location of the 137Cs source withthe temperature overlay can be seen in figure 4.7, with an obvious correlationbetween the two measurements. Figure 4.8 shows the correlation directly.In both, there is evidence that the first 8 to 12 hours of measurements do notcorrelate with temperature. It is thought that since the interior of the PP isa vacuum, the rate at which the dynodes equilibrate to the temperature isvery slow. The response of the CsI crystal itself does vary with temperature[14][16], but this only accounts for about half of the variation observed, thetotal of which was (−1.65± 0.02)%/◦C.214.3. ResultsTemperature (C)33.633.83434.234.434.634.83535.2Time and DateSep2617:00Sep2717:00Sep2817:00Sep2917:00Sep3017:00Oct0117:00Oct0217:00Peak Response / average 0.970.9750.980.9850.990.99511.0051.011.015Peak Response of Cs137 and TemperatureFigure 4.7: CsI peak response withtemperature overlay using the 137Csas the source of gamma radiation.Note that there appears to be a sta-bilization period of about 8-12h.Temperature (C)33.8 34 34.2 34.4 34.6 34.8 35 35.2Peak Response / average 0.970.980.9911.01/ ndf 2χ 1725 / 125Prob 0p0 0.007012±1.571 p1 0.000203±-0.01648   Peak Response of Cs137 as a Function of TemperatureSlopeInterceptFigure 4.8: CsI peak response as afunction of temperature. The pointsin blue mark the period over whichthe PP was stabilizing. The tempera-ture sensitivity of the combined setupwas found to be (−1.65± 0.02)%/◦C.Given the temperature correlation, it was then possible to correct thepeak locations in figure 4.7 to eliminate effects due to a variation in tem-perature. The temperature independent peak response can be seen in figure4.9. It seen that the stabilization period still exists, and that the variationin peak location has decreased. Figure 4.10 shows a histogram of the peakresponse, with a Gaussian fit to determine the variability in the responseafter corrections to temperature. This residual instability was found to be(0.28± 0.03)%.Time and DateSep2617:00Sep2717:00Sep2817:00Sep2917:00Sep3017:00Oct0117:00Oct0217:00Peak Response / average (corrected)0.9750.980.9850.990.99511.005Peak Response Corrected to 34.7 CelciusFigure 4.9: The peak response in fig-ure 4.7 corrected for temperature.projEntries 144Mean 0.0004603±0.9979 RMS 0.0003255±0.005524 Underflow 0Overflow 0 / ndf 2χ 14.33 / 17Prob 0.6435Constant 2.92±25.71 Mean 0.0003±0.9997 Sigma 0.000285±0.002846 Peak Response / average (corrected)0.975 0.98 0.985 0.99 0.995 1 1.005Number of Events0510152025Peak Response Corrected to 34.7 Celcius ProjectionFigure 4.10: A projection of the peakresponse of the temperature correctedpoints in figure 4.9. The variabil-ity after temperature corrections was(0.28± 0.03)%.224.4. Conclusion4.4 ConclusionUsing a variety of radioactive calibration sources, it was found that theresponse of the PP to energy deposited in the pure CsI crystal is linear.Gamma rays from a 137Cs source were used to track the peak response ofthe system over the course of a week. Despite the use of the incubator,some temperature variation was observed with a corresponding change inthe peak location. This change amounted to (−1.65 ± 0.02)%/◦C and issignificantly larger than the variation of the CsI itself. After correctingfor the temperature variation, the residual variability of the response was(0.28± 0.03)% from the average.23Chapter 5Hamamatsu Measurements5.1 IntroductionSixteen photopentodes (PP) were acquired from Hamamatsu, who provideda data sheet specifying several measurements performed before shipping.These measurements were the cathode luminous sensitivity (µA/lm), theanode luminous sensitivity (µA/lm), the dark current (nA), and the cath-ode blue sensitivity index. The cathode luminous sensitivity is proportionalto the quantum efficiency of the cathode, whereas the anode sensitivity isproportional to the product of the gain and the quantum efficiency. Thedark current is the current through the anode after 30 minutes sealed in adark box, whereas the cathode blue sensitivity index is the cathode luminoussensitivity with a blue filter applied to the light source. All of the Hama-matsu measurements were carried out at an operating voltage of −750 V,and a Tungsten filament lamp was used as the light source. With these mea-surements and the 16 PP, the goals for this part of the experiment were todetermine some of the tube-to-tube variation as well as to search for corre-lations between the Hamamatsu measurements and the ADC/energy slopeof chapter 4 to set parameters for quality control.The experimental setup for this chapter was unchanged from the shortterm stability, described in section Results5.2.1 On the Hamamatsu Measured ValuesOf the four measured values provided by Hamamatsu, the dark current wasthe least useful. The dark currents were measured to a precision of 0.01 nA,which was not good enough to properly get a sense of how well the PP wereperforming. The average dark current of the 16 PPs was (0.011±0.007) nA.The anode and cathode luminous sensitivity are both measures of the PPresponse to light. The distribution of each can be found in figure 5.1. It wasseen that the cathode luminous sensitivities varied by 13% across the tubes,245.2. ResultsEntries  16Mean    83.38RMS     10.73Underflow       0Overflow        0Cathode Luminous Sensitivity (uA/lm)0 20 40 60 80 100 120Number of Entries01234 Entries  16Mean   1.446e+04RMS      4323Underflow       0Overflow        0Anode Luminous Sensitivity (uA/lm)0 10000 20000 30000Number of Entries00.511.522.53Figure 5.1: Cathode and anode luminous sensitivities as measured by Hama-matsu at an operating voltage of −750 V were on average (83 ± 3) µA/lmand (14± 1) mA/lm respectively.while the anode luminous sensitivities varied by 29%. Note that these valueswere measured using a Tungsten filament lamp operated 2856 K, whichproduces light peaking around 1000 nm [26], far from the UV region. Giventhat the scintillation light from the CsI peaks at 315 nm, the Hamamatsumeasurements of the cathode and anode luminous sensitivities may not beindicative of the PP performance in the ECL.Entries  16Mean    9.412RMS    0.4092Underflow       0Overflow        0Cathode Blue Sensitivity Index0 2 4 6 8 10 12Number of Entries0246810Figure 5.2: Cathode blue sensitiv-ity index as measured by Hama-matsu at −750 V. The index hasan average of 9.4 ± 0.1 for the 16PP.The final measurement provided byHamamatsu was the cathode blue sen-sitivity index. The index is simply thecathode luminous sensitivity where theTungsten lamp had a Corning Cs 5-58blue filter at half-stock thickness fil-tering the light. This produces lightwith an emission spectrum that peaksat 420 nm [21]. Since the light hasbeen filtered, it is not appropriate touse lumens as a unit of measurement,so Hamamatsu reports the value as aunitless quantity. The distribution ofthe cathode blue sensitivity index canbe seen in figure Gain vs OperatingVoltageSince the cathode luminous sensitivity is proportional to the quantum effi-ciency, and the anode luminous sensitivity proportional to the product of255.2. Resultsthe quantum efficiency and the internal gain of the PP, the quotient of thetwo is the internal gain of the PP. However, the Hamamatsu measurementswere taken at an operating voltage of −750 V, whereas the measurementsperformed were taken at −1000 V. Additionally, since the endcap ECL willbe in a magnetic field, it is expected that the gain of the PP will dropby roughly a factor of 3.5 [27]. To simulate this effect, and to correct theHamamatsu measured gains to −1000 V, the peak locations of each of thecalibration sources were tracked as a function of the operating voltage ofthe PP. Figure 5.3 shows the ADC location of the 1.063 MeV peak from the207Bi source as a function of PP operating voltage.Voltage Input to PP (V)600 650 700 750 800 850 900 950 1000Peak - Pedestal600700800900100011001200130014001.063 MeV Peak - Pedestal / ndf 2χ 61.11 / 7Prob 9.058e-11p0 2.928±-444.9 p1 0.003868±1.775  interceptslope(ADC)Figure 5.3: To find the change in the gain, the peak location of the 1.063 MeV207Bi decay was tracked as a function of the PP operating voltage.Pictured in figure 5.4 is the distribution of the internal gains of the PPsat −1000 V, the nominal value. The average gain at −1000 V was 255± 11.The gain at any given voltage can be found by:Gain(V ) =Peak(V )Peak(−750) ·Gain(−750), (5.1)where the gain at−750 V is given by the ratio of the luminous sensitivities byHamamatsu. Figure 5.5 shows the average gain of all 16 PP as a function ofthe operating voltage, where the relationship between the operating voltageand the PP gain was seen to be linear. This was the case with all of theindividual PP as well. Since the gain of the PP depends on the secondaryemission ratio (δ) for each of the three dynodes, it was expected that δwould increase linearly with the voltage, and therefore the gain of the PP265.2. Resultswould have increased non-linearly. It is perhaps due to the unequal voltagedistribution to each of the dynodes that this was not the case.ahistEntries 16Mean 11.05±255.1 RMS 7.813±44.2 Underflow 0Overflow 0PP Gain at -1000V0 50 100 150 200 250 300 350 400 450 500Number of Entries0123456Hamamatsu Gains Converted to 1000VFigure 5.4: The gain of the PP fromthe values measured by Hamamatsu,as predicted for an operating voltageof −1000 V.Operating Voltage (-V)300 400 500 600 700 800 900 1000Average Gain050100150200250/ ndf 2χ 26 / 6−1.191eProb 1intercept 14−4.739e±79.05 −slope 17−6.876e±0.3341 Mean Gains Over All PP as a Function of HVFigure 5.5: The average gain of allPP as a function of the operatingvoltage. For each of the individualgains, see figure D. ComparisonsIt is first important to note that the cathode and anode luminous sensitivitiesare intrinsically linked: the output of the anode for a given amount of lightmust depend on the effectiveness with which the cathode produces electrons.Figure 5.6 shows that the expected correlation between these two quantitiesis indeed present.Cathode Luminous Sensitivity (uA/lm)70 75 80 85 90 95 100 105Anode Luminous Sensitivity (uA/lm)8000100001200014000160001800020000220002400026000/ ndf 2χ 2.227e+07 / 14Prob 0p0 2470±-1.785e+04 p1 29.38±387.6   interceptslopeFigure 5.6: The anode and cathodeluminous sensitivities are correlated.Anode Luminous Sensitivity (uA/lm)8000 10000 12000 14000 16000 18000 20000 22000 24000 26000Slope (bin/MeV)7008009001000110012001300140015001600/ ndf 2χ 4678 / 14Prob 0p0 3.642±428.8 p1 0.0003425±0.04306   slopeinterceptFigure 5.7: The slopes from the peakvs energy plots as a function of theanode luminous sensitivity.The slope from the peak vs energy plots, as seen in figure 4.6, is a verysimilar measurement to the anode luminous sensitivity. Consider the units275.2. Resultsof the slope: bin/MeV specifies the PP response as a function of the incidentlight, whereas the anode luminous sensitivity is a measure of the same withµA/lm. Additionally, both should depend on the sensitivity of the cathodeand the anode. Figures 5.7, and 5.8 show the expected correlations arepresent. Additionally, with the blue filter present, the expected relationshipstill holds (fig. 5.9).Cathode Luminous Sensitivity (uA/lm)70 75 80 85 90 95 100 105Slope (bin/MeV)7008009001000110012001300140015001600/ ndf 2χ 7783 / 14Prob 0p0 8.462±-71.97 p1 0.1153±12.99   interceptslopeFigure 5.8: The slopes from the peakvs energy plots as a function of thecathode luminous sensitivity.Cathode Blue Sensitivity Index8.6 8.8 9 9.2 9.4 9.6 9.8 10 10.2Slope (bin/MeV)7008009001000110012001300140015001600/ ndf 2χ 4258 / 14Prob 0p0 25.7±-2394 p1 2.841±361.9   slopeinterceptFigure 5.9: The slopes from the peakvs energy plots as a function of thecathode blue index.Since both the slopes and the anode luminous sensitivity are measuredat the anode, they should be dependent on the gain of the PP. Figures 5.10and 5.11 show that this is indeed the case.The cathode luminous sensitivity is proportional to the quantum effi-ciency. The anode luminous sensitivity is essentially the same as the cath-ode luminous sensitivity, but with the added factor of the gain. Thereforetheir quotient, the gain, should be independent of the quantum efficiency.This was not the case, as seen in figure 5.12. Furthermore, the componentleftover from dividing the slopes by the gain should be correlated with thecathode luminous sensitivity. As seen in figure 5.13, this was also not thecase.285.2. ResultsHamamatsu Gain Corrected to -1000V180 200 220 240 260 280 300 320 340 360 380Slope (bin/MeV)7008009001000110012001300140015001600/ ndf 2χ 483.1 / 14Prob 0intercept 16.04±45.5 −slope 0.06985±4.334 Figure 5.10: The slopes from thepeak vs energy plots are correlatedwith the Hamamatsu measured gains,corrected to the operating voltage of−1000 V. This was the voltage atwhich the peak locations were mea-sured.Hamamatsu Gain Measured at -750V120 140 160 180 200 220 240Anode Luminous Sensitivity (uA/lm)8000100001200014000160001800020000220002400026000 / ndf 2χ 1.046e+07 / 14Prob 0intercept 1297±1.067e+04 −slope 7.518±147.8 Figure 5.11: The anode luminoussensitivity as a function of the gainmeasured by Hamamatsu.Hamamatsu Gain at -750V (anode/cathode)120 140 160 180 200 220 240Cathode Luminous Sensitivity (uA/lm)707580859095100105/ ndf 2χ 322 / 14Prob 0intercept 7.197±25.69 slope 0.04172±0.3391 Figure 5.12: The cathode luminoussensitivity as a function of the gainwas seen to be correlated.Cathode Luminous Sensitivity (uA/lm)70 75 80 85 90 95 100 105Slope/Gain (bin/MeV)3.63.844.24.4Figure 5.13: The quotient of theslopes from the peak vs energy plotsand the Hamamatsu measured gaincorrected to −1000 V, as function ofthe cathode luminous sensitivity.It was observed that the slope/gain was not correlated with any other ofthe measured parameters, indicating that the variance in experimental setupmay have increased the uncertainty of the measurement beyond expectedlevels. It should be emphasized that the slope/gain should be correlatedwith the cathode luminous sensitivity, as the slope is very similar to theanode luminous sensitivity and the anode sensitivity divided by the gain is295.3. Conclusionalgebraically identical to the cathode sensitivity.The quotient of the width and the peak location (the resolution), isa function of the energy deposited in the crystal, as seen in figure 5.14.Therefore, high energy events should expect a low resolution, less than 8%for signals above 2 MeV. The resolution was also found to be independentof the PP operating voltage.Energy (MeV)0.4 0.6 0.8 1 1.2 1.4 1.6 1.8Resolution0. Resolution (no errors)Figure 5.14: Average resolution as a function of the energy. As expected,the resolution decreases with increasing energy deposit in the CsI.5.3 ConclusionThe Hamamatsu measured quantities were compared with the slope from thepeak response vs energy plots. While correlations were observed between theslope and the anode luminous sensitivity, the cathode luminous sensitivity,and the cathode blue index, the experimental precision was not great enoughto determine if a correlation was present with the PP gain factored out of theslope. There was also a correlation between the cathode luminous sensitivityand the gain.Table 5.1 provides a summary of the average quantities measured for the16 PP.Furthermore, the gain was found to be linear with the operating voltage.This was an unexpected result, as a non-linear relationship was expectedfrom the increased secondary emission ratios from the dynodes. It was alsoseen that the resolution improves with increased energy deposit in the CsIcrystal.305.3. ConclusionTable 5.1: Average Hamamatsu quantities which were measured at −750 Vand the PP internal gain at −1000 V operating voltage for a population of16 PP.Quantity Average Variation Among PPAnode Luminous Sensitivity (14± 1) mA/lm 29%Cathode Luminous Sensitivity (83± 3) µA/lm 13%Cathode Blue Index (9.4± 0.1) 4%Dark Current (0.011± 0.007) nA 64%Gain at −1000 V (255± 11) 17%31Chapter 6Excess Noise Factor6.1 IntroductionThe excess noise factor is defined as the quotient of the squares of the reso-lutions at the cathode and the anode, as given in the following equation:(σcNc)2F =(σaNa)2, (6.1)where F is the excess noise factor, Nc,a is the number of electrons at thephotocathode and anode respectively, and σc,a are the corresponding widthsof the distributions [21]. The excess noise factor arises because of the mul-tiplication process in the photopentode (PP), as described in chapter 2.Taking advantage of the Poisson distributed nature of the photoelectrons,this can be re-written in the following linear form:σ2m = F ·Nc + σ2o , (6.2)where σm is the width of the signal in units of number of photoelectronsand σo is the electronic noise. For details on the algebraic manipulation seeappendix C.2. Recall that an electron at the cathode is named a photoelec-tron, and therefore the goal of this measurement was to extract the excessnoise factor by observing the width of the output distribution as a functionof the number of photoelectrons.6.2 Experimental SetupOriginally, the idea was to use a standard PMT (appendix B.2.3) to countthe number of incident photons, and to use the width of the signal producedby the PP to determine the excess noise. This idea was discarded because itwas not possible to isolate the quantum efficiencies of the two phototubes,both of which are unknown quantities. It was only possible to find theirratio (appendix C.4), which could only be solved for in such a way thatusing the standard PMT was redundant. A full justification of why the326.2. Experimental SetupFigure 6.1: Excess noise setup top view. This setup was encased in the darkbox with the laser light being transmitted via fibre optic.standard PMT was not useful can be found in appendix C.3. Despite this,even the latest iteration of the setup incorporated this idea into its design.As seen in figure 6.1, the setup consisted of four parts: the phototubes, thescreen, the fibre optic, and the support structure. The PMT and the PPwere run at the operating voltages of −2200 V and −1000 V respectively.The laser was the source of UV light in this measurement. It producedvariable intensity pulsed UV light of 405 nm at an adjustable rate (appendixB.2.4). In this case, the pulse rate used was 350 Hz. The laser itself waskept outside of the dark box, and the light was transmitted into the box byfibre optic and optical feedthrough. The use of the laser was motivated bythe ease at which one could adjust the light intensities, as well as the rangeof intensities available.To ensure that the light was uniform across the face of both phototubes,a diffuse screen was used to reflect the laser light (appendix B.2.5). Thescreen was held perpendicular to the phototubes, and the phototube toscreen distance could be changed to adjust the light intensity, in additionto varying the laser voltage. This was needed because the laser requiredthat the voltage be set beyond a threshold, effectively setting a minimumnonzero intensity for the light produced. To further lower the light intensity,the screen was moved further away from the phototubes.The support structure housed all the components. The purpose of thisdevice was to ensure that the screen, PP, PMT, and laser were all securedin place with respect to each other. The screen was allowed to slide withina track, allowing the distance between the screen and the phototubes tochange while keeping the screen perpendicular to the plane of the faces ofthe phototubes. The support structure also had a port in which the fibreoptic was installed, and could be rotated to adjust the position of the lighton the screen. The two phototubes had custom built supports to secure the336.3. ResultsFigure 6.2: Excess noise setup front view. The PP and the standard PMTwere held in parallel so that they received the same amount of light. Theangle at which the laser was held relative to the screen was adjustable.phototube in place, and each could be exchanged in a reproducible manner.Upon exchange, the angle at which the phototube was held did not change,nor would the distance between the phototube and the screen. Figure 6.2shows the portion of the support structure that housed the phototubes andthe fibre optic.The DAQ for this measurement again made use of the peak sensingLeCroy L2259B ADC, and the MIDAS software. The ADC gate was pro-vided by a NIM signal that was coincident with the TTL signal sent totrigger the laser. Since the standard PMT did not need a preamp or shaper,the ADC used to measure the output was an integrating ADC, the LeCroyL2249. Both were triggered simultaneously. The output of the standardPMT was attenuated by a factor of 0.3, to ensure that the output signalwas within the range of the L2249 ADC. With similar reasoning, the outputof the Ortec 474 shaper was amplified by a factor of 10. For more detailsplease refer to appendix A.3 for the circuit diagrams.6.3 Results6.3.1 PMT Single Photon MeasurementDuring the time it still seemed likely that the standard PMT would beused, it was instructive to determine the gain of the standard PMT. A small346.3. Resultslight leak was introduced in the dark box by leaving the optical feedthroughdisconnected from the fibre optics. Using a random trigger, single photonscould be found by hand on a Tektronix Oscilloscope. A series of singlephoton waveforms were saved and were then integrated in ROOT. With the50 Ω input resistance of the scope and cables, the equivalent charge for eachwaveform was calculated and histogrammed, as seen in figure 6.3. Dividingthe histogram mean by the charge of an electron, indicates that the gain ofthe R5113-02 PMT was (3.99± 0.15)× 106 with the attenuation.intHistEntries 165Mean 14−2.351e±13 −6.391eRMS 14−1.662e±13 −3.02eUnderflow 0Overflow 0Charge (C)0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.612−10×Number of Entries0510152025Figure 6.3: Single photon measurements made with a 0.3 attenuationon the output. The PMT gain, including attenuation, was found to be(3.99± 0.15)× Time to StabilizationIt was observed that the signal from the PP varied significantly when thelaser was first turned on, prompting an investigation as to the length oftime until the PP signal stabilized. For this measurement, the conditionsarising from switching the phototubes or adjusting the screen position weremimicked: both the laser and the phototube operating voltage sources wereturned off, and the dark box was left open for about 15 minutes. The darkbox was then closed, and both power sources were turned back on. Theamplitudes of the PP signal were then immediately recorded in 120 secondsegments over the next hour. An example of the resulting spectrum from onesuch segment can be seen in figure 6.4, with Novosibirsk fit. The Novosibirskfunction was chosen because the distribution was found to be asymmetric.356.3. Resultsrun00000388Entries  104278Mean    0.209±  341.9 RMS    0.1478±  67.49 Underflow       0Overflow        0 / ndf 2χ   521 / 499Prob   0.2397norm      2.4± 617.2 peak      0.3±   341 width     0.15± 67.06 eta       0.00273±0.01109 − ADC0 200 400 600 800 1000 1200 1400 1600 1800 2000Number of Entries0100200300400500600700PP Raw Output From LaserFigure 6.4: Peak height spectrum from pulsed laser as measured by the PP,with Novosibirsk fit. The peak location and width of this fit are used todetermine the excess noise factor.See appendix D.5 for a comparison with a Gaussian fit.It was found that after about 15 minutes, the variation in the amplitudeof the PP signal was less than about 0.5% and that this was no longer asignificant source of uncertainty in the final measurement. The stability ofthe system can be seen in figure 6.5, a histogram of the same value foundin figure 6.6. After fitting the normalized histogram with a Gaussian, thestability was measured to be (0.38± 0.06)% over the course of an hour.366.3. ResultsTime Elapsed (s)0 500 1000 1500 2000 2500 3000 3500Peak Location/First Peak Location0.9750.980.9850.990.9951Figure 6.5: Time to stabilizationfor the PP and laser setup. Thecurve was fitted with the sum of anexponential and a constant to helpdetermine the ideal point at whichthe PP could be considered stabilized(15 minutes).peakHistEntries 90Mean 0.0008306±0.9591 RMS 0.0005873±0.007879 Underflow 0Overflow 0/ ndf 2χ 24.38 / 11Prob 0.01124Constant 3.17±17.82 Mean 0.0005±0.9565 Sigma 0.000590±0.003835 Peak Location/First Peak0.95 0.955 0.96 0.965 0.97 0.975 0.98 0.985 0.99 0.995 1Number of Entries0510152025Figure 6.6: The peaks from theNovosibirsk fits over the course ofan hour. Here, the histogram hasbeen fitted with a Gaussian, to showthat the variability in pulse height is(0.38± 0.06)%.6.3.3 Uncertainty of Measurementpos2HistCH0Entries  9Mean    6.726±  722.9 RMS     4.756±  20.18 Underflow       0Overflow        0Peak Location (ADC)550 600 650 700 750 800 850Number of Entries00.511.522.533.54Peak Locations for R11283pos1HistCH0Entries  9Mean    9.036±  661.3 RMS     6.389±  27.11 Underflow       0Overflow        0pos1pos2Figure 6.7: The reproducibility of thePP measurement after exchanging thephototubes.To determine the reproducibility ofthe setup, the phototubes were ex-changed several times, each timetaking a measurement of the inci-dent light. Each run had the samelight intensity setting, and allowedthe setup to stabilize for the estab-lished period before a measurement.The uncertainty, as shown in figure6.7, was 4% in one position and 3%in the other position. Despite thelack of phototube exchange in thefinal measurement, this should betaken as an estimate of the uncertainty in incident light as the PPs werecycled through.6.3.4 Calibration Pulse MeasurementTo convert the MIDAS signal to the number of photoelectrons, it was firstnecessary to convert the bins read from the ADC into the amount of chargeat the anode of the PP. This was done with the aid of the calibration test376.3. Resultspulse input of the preamp. The test pulse allowed for the injection of agiven amount of charge into the preamp to simulate a real photon event.The mechanism for the injection was a (1± 0.25) pF capacitor [23]. A pulsegenerator was used to send square waves of width 127 µs into the capacitor,and the capacitor converted the input voltage into a pulse of correspondingcharge. A screenshot of the output calibration pulse after the shaper andx10 amplification can be seen in figure D.10.The input charge to the preamp was then varied and the correspondingADC signal observed, as seen in figure 6.8. The pedestal used here was thesame as was found with the energy linearity measurements in chapter 4. Itwas found that the conversion rate from the ADC bin to charge at the PPanode was (8.54± 0.03± 1.3)× 10−18 coulombs / bin.Calibration Pulse Input Charge (C)4 6 8 10 12 1415−10×ADC-Pedestal4006008001000120014001600 / ndf 2χ  16.3 / 6Prob   0.01223intercept  2.827±17.18 − slope     4.539e+14± 1.171e+17 Preamp to ADC CalibrationFigure 6.8: Calibration pulse results, with a linear fit. This allowsfor the conversion from ADC bin to charge at the anode of the PP:(8.54± 0.03± 1.3)× 10−18 coulombs / bin.6.3.5 Excess Noise FactorAs was outlined in section 6.1, the goal of this chapter is to measure thewidth of the signal and to plot this as a function of the signal peak, in unitsof number of photoelectrons. To this end, recall thatσ2m = F ·Nc + σ2o , (6.2)where σm is the width of the signal measured, Nc is the signal size at thecathode, and σo is the electronic noise, and all are in units of number of386.3. Resultsphotoelectrons. The strategy employed was to change the incident lightintensity by varying the screen to PP distance and to record the resultingsignal height distribution. This spectrum was fitted with a Novosibirsk func-tion, and the peak height and width for each light intensity were recorded inADC counts. The pedestal from chapter 4 was subtracted. The calibrationpulse measurement described in section 6.3.4 was used to convert the ADCvalue to charge at the anode. The anode charge was then divided by thegain and the charge of an electron to get the number of photoelectrons. Theresult of the above operations on both the width and the peak locations canbe seen in figure 6.9.Number of Photoelectrons60 80 100 120 140 160 180Width Squared (Number of Photoelectrons)150200250300350 / ndf 2χ 0.6625 / 2Prob   0.718intercept  3.833± 30.66 slope     0.0421± 1.767 Excess Noise Factor Calculations for Photopentode 7Figure 6.9: Excess noise factor cal-culation for a single PP. The ex-cess noise factor for this PP was1.77 ± 0.04, and the electronic noisewas 5.5± 0.3 photoelectrons.fHistPPEntries  16Mean   0.02589±  1.939 RMS    0.01831± 0.1036 Underflow       0Overflow        0Excess Noise Factor1.6 1.8 2 2.2 2.4 2.6 2.8 3Number of Entries012345Excess Noise Factor for 16 PP (PP Npe)Figure 6.10: Histogram of all the ex-cess noise factors of the acquired PP.The mean excess noise factor was1.9± 0.1.histEntries  16Mean    33.37±   1730 RMS     23.59±  133.5 Underflow       0Overflow        0Electronic Noise (Num Electrons at Anode)0 200 400 600 800 1000 1200 1400 1600 1800 2000Number Of Entries0123456Figure 6.11: Histogram of all the elec-tronic noise values (should be indepen-dent of PP), in units of number ofelectrons at the anode. There is 7.7%variability in the measurement.The fit in figure 6.9 is simplyequation 6.2, and therefore the slopeis the excess noise factor and theintercept is the square of the elec-tronic noise. The excess noise fac-tors for all of the 16 PP can be foundin figure 6.10, and the electronicnoise in figure 6.11. The average ex-cess noise factor from all the PP was(1.9±0.1±0.4), with the systematicerror arising from the uncertainty inthe preamp test pulse. The aver-age electronic noise was found to be1730± 33 electrons at the anode, or396.3. Resultsequivalently (32± 1) keV from the Saint-Gobain CsI crystal.histEntries  16Mean   0.001128± 0.03183 RMS    0.0007976± 0.004512 Underflow       0Overflow        0Equivalent Noise Energy (MeV)0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1Number Of Entries01234567Figure 6.12: Histogram of all the electronic noise values, in units of equiva-lent energy from the Saint-Gobain CsI crystal. Note that this value is muchsmaller than the noise found in chapter 3, showing the difference in crystalquality.6.3.6 Excess Noise at Reduced Operating VoltageOperating Voltage (-V)300 320 340 360 380 400Excess Noise Factor22. Excess Noise Factor HV Dependence (PP Npe)Figure 6.13: Excess noise factor as afunction of operating voltage, at lessthan 1/3 of the maximum gain. Seefigure 5.5 for details on the gain.Since the ECL is in a magnetic field,it is expected that the gain will dropby a factor of about 3.5 [27]. To sim-ulate this, the operating voltage ofthe PP was reduced and the excessnoise factor measurement discussedin the previous section was repeatedfor each voltage setting (fig. 6.13).The excess noise factor was found toincrease for operating voltages be-low −400 V, or a gain of about 55.Above this, there was no significantrelationship between the operatingvoltage and the excess noise factor.406.4. Conclusion6.4 ConclusionThe average excess noise factor for the R11283 Hamamatsu photopentodewas found to be (1.9± 0.1± 0.4) at a gain of 255 (−1000 V), and increasingto 3.5± 0.1 at a gain of 25 (−300 V). The large systematic error arose fromthe 25% uncertainty in the preamp test input capacitor [23]. The trend seenin the PP is the inverse of that seen in avalanche photodiodes (APD), wherethe excess noise factor increases with gain [28]. The PP starts to becomecompetitive with the cited APDs at a gain larger than ≈ 50 (−350 V).Given that the gain of the particular PP tested would drop to about 105in the magnetic field, it is expected that the performance will be better aswell. Note also that the PP holds up when in direct competition with theHamamatsu S8664-1010(S8664-55) APDs for use in the Belle II ECL, wherethe excess noise factor was found to be 3.4± 0.4(5.1± 0.5) [29].The electronic noise of the preamp was also found in units of number ofelectrons at the anode, 1730±33. This is comparable to the value measuredby the Universite´ de Montre´al, 1500 e− [23]. The equivalent noise energy wasalso be found from this measurement, and was reported as (32± 1) keV forthe polished pure CsI crystal provided by Saint-Gobain; less than half of thevalue found for the large BINP CsI crystal in chapter 3. This discrepancy islikely due to the difference in crystal quality, size, and polish. It should benoted that measuring the RMS of the spectrum of waveform heights is notsufficient to make a measurement of the electronic noise, since the excessnoise factor is not taken into account.41Chapter 7Aging and Lifetime7.1 IntroductionFigure 7.1: PP array encased in incu-bator with glued CsI pucks.The primary result needed from thephotopentode (PP), as outlined inthe Belle Technical Report [6], wasthe lifetime of the PP. To do this,UV light comparable to the dom-inant wavelength of the pure CsIemission spectrum was flashed on aarray of 16 PP (fig. 7.1) for about48 days. A 207Bi source was usedto track the performance of the PP.Approximately 70 years worth ofBelle II operation was simulated inthis manner, and the change in PPperformance and stability was mea-sured.The predicted maximum radia-tion dosage rate in the ECL is about1.3Gy/yr [30], and the large Belle IIBINP crystal has a light yield of 185 photoelectrons per MeV [31]. For 7years of operation there should be about (2.84×1014) MeV deposited in thecrystal. With the PP internal gain assumed to be 2503 in the~B field, thiscorresponds to 0.701 C at the anode. As a safety factor, a total charge anorder of magnitude larger was induced in the anodes.427.2. Experimental Setup7.2 Experimental SetupIt was thought that the amount of current passed by the anode of the PPwas the primary cause of the PP aging. An array of 16 PP was built tostudy the effects of aging. The array was read out using the V4 preamp,in conjunction with the Universite´ de Montre´al shapers (appendix B.1.2).Each PP had its own two inch diameter pure CsI puck provided by Amcrysglued to its face using a silicone rubber glue (appendix B.2.2). The PP arraywas encased in an incubator set to keep the temperature at roughly 35 ◦C(appendix B.3.5). The incubator was outfitted with a UVT acrylic door,and N2 was pumped into the top of the incubator to help keep the humiditylow. The humidity and temperature sensor was placed inside the incubator,along one of the walls.To accelerate the aging process, an LED flasher provided by the Univer-sity of McGill (appendix B.2.6) was used to induce a large current throughthe PP dynodes and anode. Figure 7.2 shows the LED flasher and incubatorsetup. The flasher was set to a distance of about 40 cm from the PP arrayand was run with a DC power source set to induce about 2 µA of currentthrough the anode. The current at the anode was measured directly on oneof the PP by means of a modified preamp (appendix B.1.3). The currentwas read out by the Keithley 6485 Picoammeter, and the result recordedwith MIDAS. To track the stability of the PP, a 207Bi source was used. Thiswas taped to the mechanical support structure holding the LED array. As(a) LED Flasher with 207Bisource attached.(b) LED Flasher and PP array in incubator.Figure 7.2: Setup for aging the PP in the dark box.437.3. Resultsmentioned, the gain of the PP array was kept at a third of its maximumvalue by setting the operating voltage to 491 V, with the exception of thetwo lowest gain PP where the operating voltage was set to 618 V. This wasdone to ensure that the peak locations remained within the range of theADC while the PP was aged. The average gain of the 16 PP at 491 V was85± 3. Refer to appendix A.4 for the associated circuit diagram.7.3 Results7.3.1 Current Baseline and EstimationTo estimate the current in each of the PP while the aging was ongoing, thecurrent of each PP was first measured under identical conditions. Each PPwas done in sequence, with three repetitions of the sequence. Since only oneof the PP had the current measured throughout the entire aging process,the currents from the baseline measurement were compared to provide anestimate of the current passing through the anode of each PP. Equation 7.1describes the estimated current in the jth PP (Ij), given the currents fromthe baseline measurement (I(B)j ):Ij =I(B)jI(B)rIr, (7.1)where Ir is the current measured from the reference PP with the modifiedpreamp. Given that not all of the PP were run at the same operatingvoltage, the relationship between the current and the operating voltage wasestablished and can be found figure B. Aging and LifetimeTime Elapsed (days)0 10 20 30 40 50Charge (C)0246Total Charge Passed Through Anode in PP10Figure 7.3: The total charge passedthrough the anode of the current mon-itoring PP.Figure 7.3 shows the cumulativecharge passing through the anodeof the monitoring PP as a functionof time. This curve is the result ofintegrating the current with respectto time. The raw current measure-ment can be seen in figure D.11 inthe appendix. Approximately 7 C ofcharge was passed through the an-ode over 48 days.447.3. ResultsFigure 7.4 shows the spectra of the LED aging light pulses. Since theLED array was run with a DC current, the DAQ recorded a range of smallpulses at a rate of about 1 MHz. This is ideal for the aging as a large num-ber of low energy events is the expected form of the background in the BelleII ECL. The 207Bi spectrum in figure 7.5 was fitted with the sum of twoNovosibirsk functions and an exponential. The exponential roughly modelsthe backscattering and background from the other peaks, while the Novosi-birsk captures the location of the peaks. It is the peaks of the Novosibirsksthat was used to track the performance of the PP.L2249_15_7Entries  2388266Mean   0.1068±  276.4 RMS    0.0755±    165 Underflow       0Overflow        0ADC0 200 400 600 800 1000 1200 1400 1600 1800 2000Number of Entries0100020003000400050006000700080009000PP16 LED Intensity DistributionFigure 7.4: Distribution of the pulseheights from the LED used to age tothe PP.ADC100 200 300 400 500 600 700Number of Entries0200400600800100012001400PP01 on 2015-7-30 at 17:14Figure 7.5: Spectrum of the pulseheights due to the 207Bi source, withfit. The peak locations were used totrack the PP performance.Energy Deposited (KeV)600 800 1000 1200 1400 1600 1800Peak Location (ADC)2003004005006007008009001000 / ndf 2χ 1.862 / 1Prob   0.1724intercept  0.9513±123.1 − slope     0.001374± 0.6205 PP01Figure 7.6: Typical 207Bi peak lo-cations as a function of energy.The spectrum was measured over thecourse of a three hour period in orderto get the statistics necessary to seethe third peak.There was a pedestal in the mea-surements, where the projected out-put of the ADC for zero energy wasnon-zero, which was due to a DCoffset in the shaper output. To findthe offset for each of the PP, a threehour measurement of the 207Bi spec-trum was taken and all three peakswere fitted with the sum of threeNovosibirsk functions and an expo-nential. The increased time of therun was needed to accurately seethe highest energy decay (appendixD.7). The peak locations were thenplotted as a function of the energydeposited in the crystal (fig. 7.6),457.3. Resultsand the intercept of the linear fit was then subtracted from all the measuredpeak locations during the aging process.Figure 7.7 displays the temperature and humidity reading from insidethe incubator. It was noted that the temperature was held within about2 ◦C and the humidity within about 5% relative humidity, with allowancesfor the system to stabilize at the beginning. Because the system was not inthermal equilibrium, the first measured 207Bi spectrum was discarded, as itwas taken within this time.Days Elapsed0 10 20 30 40 50Temperature (C)3032343638(a) TemperatureDays Elapsed0 10 20 30 40 50Relative Humidity (%RH)152025303540(b) HumidityFigure 7.7: Temperature and humidity were recorded in the incubator. Notethat this does not reflect the conditions outside of the incubator, where theLED flasher was placed.Figure 7.8 shows the temperature dependence of the 207Bi peaks of thecontrol PP, with linear fit. It was seen that the effects due to temperaturewere not large, with most corrections to the peak location at the level of1-2%, after pedestal subtraction. It was found that the temperature depen-dence of the PP did not vary much from tube to tube, and so each peakof every PP was corrected by the corresponding peak from the control PP,after pedestal subtraction.Using the 207Bi source and the Amcrys CsI crystals, the 0.57 MeV peakvaried by (−1.3±0.6)%/◦C and the 1.064 MeV peak varied by (−1.2±0.4)%/◦C.In comparison, the 137Cs source and St. Gobain CsI crystal from chapter4.3 produced a temperature variation of (−1.65± 0.02)%/◦C.Figure 7.9 compares the product of the gain and the quantum efficiencybetween the control PP and an aged PP.467.3. ResultsTemperature (C)36.4 36.6 36.8 37 37.2Relative Peak 1 Location0.99511.0051.011.015PP03(a) The 0.57 MeV peak varied by(−1.3± 0.6)%/◦C.Temperature (C)36.4 36.6 36.8 37 37.2Relative Peak 2 Location0.99511.0051.01PP03(b) The 1.064 MeV peak varied by(−1.2± 0.4)%/◦C.Figure 7.8: Variation of the 207Bi peaks of the control PP with temperature,after the aging was completed.Charge Through Anode (C)0 2 4 6Relative Gain X QE0.90.951PP03(a) The control photopentode (PP03).Charge Through Anode (C)0 2 4 6 8Relative Gain X QE0.90.951PP16(b) An aged photopentode (PP16).Figure 7.9: The slope of the peak location as a function of energy, relativeto the first measurement. In the case of the control PP, the current is thatwhich would have been flowing through the anode, if the PP was not capped.Recall that the gain × quantum efficiency comes from the slope of thepeak location vs energy plots. In this case, the slope comes solely from thefirst two peaks, as there was not enough statistics in the one hour runs todistinguish the third peak. While this aged PP’s performance decreasedby about 10%, it was seen that the control PP’s performance improved byabout 5% before returning to its initial state. The reader may have observeda plateau in the cumulative charge at about day 31 (fig. 7.3). When thecharge is converted to the charge in the control PP, this plateau correspondsto the drop in the control PP’s performance at about 3.5 C. At this time,there was a severe storm which knocked out the power at TRIUMF, and the477.3. Resultsexperiment was left without power for about 2 days. It is unknown as towhy this would have affected the control PP in this way. There may be somebenefit to studying the change in the response when a delay in operationis introduced. In contrast, the storm did not seem to affect the aged PPoverly much, suggesting that the change in performance due to aging is real.Figure 7.9b is not typical of an aged PP.As seen in figures 7.10 & 7.11, there seemed to be four main trends inthe PP aging:1. Performance that aged quickly (a burn-in), and then stabilized to aconstant level of performance (green borders in figures 7.10 & 7.11).2. Performance that had the burn-in period, but never stabilized. Typ-ically, this trend ended with what appears to be a linear dependenceon the charge (blue).3. Performance that decreased steadily throughout the aging process anddid not show the burn-in period (black).4. Performance that did not change, or increased, throughout the agingprocess (red).487.3. ResultsCharge Through Anode (C)0 2 4 6 8Relative Gain X QE0.90.951PP01Charge Through Anode (C)0 2 4 6 8Relative Gain X QE0.90.951PP02Charge Through Anode (C)0 2 4 6Relative Gain X QE0.90.951PP03Charge Through Anode (C)0 2 4 6Relative Gain X QE0.90.951PP04Charge Through Anode (C)0 2 4 6 8Relative Gain X QE0.90.951PP05Charge Through Anode (C)0 2 4 6 8Relative Gain X QE0.90.951PP06Charge Through Anode (C)0 2 4 6 8Relative Gain X QE0.90.951PP07Charge Through Anode (C)0 2 4 6Relative Gain X QE0.90.951PP08Figure 7.10: The slopes of the peak vs energy deposited for only the first twopeaks, relative to the first measurement, as a function of the charge passedthrough the anode. PP03 is the control PP. These results have been correctedfor variations in temperature.497.3. ResultsCharge Through Anode (C)0 2 4 6Relative Gain X QE0.90.951PP09Charge Through Anode (C)0 2 4 6 8 10Relative Gain X QE0.90.951PP11Charge Through Anode (C)0 2 4 6 8Relative Gain X QE0.90.951PP12Charge Through Anode (C)0 2 4 6Relative Gain X QE0.90.951PP13Charge Through Anode (C)0 2 4 6Relative Gain X QE0.90.951PP14Charge Through Anode (C)0 2 4 6Relative Gain X QE0.90.951PP15Charge Through Anode (C)0 2 4 6 8Relative Gain X QE0.90.951PP16Figure 7.11: The PP have been divided into four categories: aged for aperiod, then stopped (solid); aged, but didn’t stop (−−−); aged at a constantrate throughout (· · ·); and no aging at all (−·−). PP10 is the reference PP,with the modified preamp to measure current.507.3. ResultsFor additional details on the categorization of the PP, refer to appendixD.7. It should be noted that only PP07 had a significantly different historyfrom the rest: this was the PP that was used to set up the excess noisefactor measurements, and was used extensively with the pulsed UV laser fora few months. For the PP that experienced the burn-in, most of the agingoccurred in the first coulomb of current having passed through the anode.It should also be noted that the pedestals which can be acquired from thefirst two peaks varied by about 5% prior to the pedestal subtraction, andwas not correlated with the gain × quantum efficiency.On average, the gain × quantum efficiency decreased to (93±3)% of theoriginal performance. This is seen the following section, where the stabilityof the PPs was measured after the aging was finished. Figure 7.12 showsthe average of the last 10 measurements after the aging was completed forall PP.Average Final Relative Gain0 0.2 0.4 0.6 0.8 1Number of Entries012345Figure 7.12: Distribution of the relative gain × quantum efficiency of eachPP averaged over the last 10 measurements. The full spectrum of measure-ments can be seen in figures 7.13 & 7.14. This distribution has an averageof (93± 3)%.517.3. Results7.3.3 Post-Aging StabilityAfter the aging process, the 207Bi spectra continued to be measured at reg-ular intervals over the course of about 10 days, but the LED array remainedoff. The measurements with no light source can be seen appended the ver-tical dotted line in figures 7.13 & 7.14.PP02 appears to have some instability present, as there was about a10% change in the performance after the LEDs was turned off. It was alsoseen that perhaps some of the aging is not due to the LED as PPs 1, 4, 7,and 9 all appear to continue aging with no light source. The rest of the PPappear to have a more expected behaviour, with the performance stabilizingat the aged level and staying mostly constant. It is worth noting that withthe post-aging measurements, the control PP effectively experienced zerochange in performance.527.3. ResultsDays Elapsed0 20 40 60Relative Gain X QE0.90.951PP01Days Elapsed0 20 40 60Relative Gain X QE0.90.951PP02Days Elapsed0 20 40 60Relative Gain X QE0.90.951PP03Days Elapsed0 20 40 60Relative Gain X QE0.90.951PP04Days Elapsed0 20 40 60Relative Gain X QE0.90.951PP05Days Elapsed0 20 40 60Relative Gain X QE0.90.951PP06Days Elapsed0 20 40 60Relative Gain X QE0.90.951PP07Days Elapsed0 20 40 60Relative Gain X QE0.90.951PP08Figure 7.13: Relative gain × quantum efficiency as a function of time elapsedin the laboratory. These results have been corrected for temperature varia-tions.537.3. ResultsDays Elapsed0 20 40 60Relative Gain X QE0.90.951PP09Days Elapsed0 20 40 60Relative Gain X QE0.90.951PP11Days Elapsed0 20 40 60Relative Gain X QE0.90.951PP12Days Elapsed0 20 40 60Relative Gain X QE0.90.951PP13Days Elapsed0 20 40 60Relative Gain X QE0.90.951PP14Days Elapsed0 20 40 60Relative Gain X QE0.90.951PP15Days Elapsed0 20 40 60Relative Gain X QE0.90.951PP16Figure 7.14: The LEDs were switched off after 48 days, marked by the ver-tical dotted line. PP10 is the reference PP, with the modified preamp tomeasure current.547.3. ResultsRelative Gain X QE0.92 0.94 0.96 0.98 1Number of Entries051015PP01Relative Gain X QE0.92 0.94 0.96 0.98 1Number of Entries051015PP02Relative Gain X QE0.92 0.94 0.96 0.98 1Number of Entries051015PP03Relative Gain X QE0.92 0.94 0.96 0.98 1Number of Entries051015PP04Relative Gain X QE0.92 0.94 0.96 0.98 1Number of Entries051015PP05Relative Gain X QE0.92 0.94 0.96 0.98 1Number of Entries051015PP06Relative Gain X QE0.92 0.94 0.96 0.98 1Number of Entries051015PP07Relative Gain X QE0.92 0.94 0.96 0.98 1Number of Entries051015PP08Relative Gain X QE0.92 0.94 0.96 0.98 1Number of Entries051015PP09Relative Gain X QE0.92 0.94 0.96 0.98 1Number of Entries051015PP11Relative Gain X QE0.92 0.94 0.96 0.98 1Number of Entries051015PP12Relative Gain X QE0.92 0.94 0.96 0.98 1Number of Entries051015PP13Relative Gain X QE0.92 0.94 0.96 0.98 1Number of Entries051015PP14Relative Gain X QE0.92 0.94 0.96 0.98 1Number of Entries051015PP15Relative Gain X QE0.92 0.94 0.96 0.98 1Number of Entries051015PP16Figure 7.15: Distribution of the gain × quantum efficiencies for all thePP after the aging was ended. PP10 is the reference PP with the modifiedpreamp, and PP03 was the control PP that was not aged.RMS/Mean0 0.01 0.02 0.03 0.04 0.05 0.06Number of Entries00.511.522.53Figure 7.16: Distribution of theRMS/mean of each PP (projection offigures 7.13 & 7.14 onto the y-axis).This distribution has an average of(2.19± 0.26)%.Figure 7.15 shows the distribu-tion of the product of the gain andthe quantum efficiency after thenaging was ended. Figure 7.16 showsthe RMS of the gain × quantum ef-ficiencies seen in figures 7.13 & 7.14as a measure of the variability ofthe system, including the aging. Itwas found that the average relativeRMS/mean was (2.19± 0.26)%.Comparing the variability inthe control tube during the aging,the control PP had a variability(RMS/mean) of (1.2 ± 0.1)%, andafter the aging was over the vari-ability was (0.27 ± 0.04)%. It is possible that this variability is simply dueto the decreased run time post-aging, however this is also probably more557.4. Conclusionrepresentative of the variance of the PP signal.7.4 ConclusionSimulating usage in a 1.5 T axial magnetic field, the PP were run at anoperating voltage that set the PP gain to about one third of the typicalmaximum [27], and were aged with a UV light source. The aging processtook place over 48 days, with approximately 7 C of charge passing throughthe anode, equivalent to 70 years of standard Belle II operation. Over thisperiod, the performance of the 14 tested PP was reduced to (93 ± 3)% ofthe initial value. For some, most of this aging occurred during a burn-inperiod lasting for about 1 C, but there remains a large scatter of differentbehaviours among the PP. The PP performance was tracked with the use ofa 207Bi calibration source. With two constant visible energy deposits in thecrystal, the product of the gain and quantum efficiency was found and usedas a measure of the PP performance. This quantity dictates the outputsignal size for a given incident light intensity. There is evidence that theaging continued past the point where the light source was switched off. Itis possible that the PP aging is not a function of the charge passed throughthe anode, but rather of some other quantity. Alternatively, there could bea delay in the aging, where the effects of the light source are offset by adelay in real time. The excess noise factor measurement, which made useof the same set of PP, was chronologically prior to the aging measurementand some burn in may have occurred then. In the measurements presented,there is no evidence of a delay at the start of the measurement. It should benoted that the control PP, which was not exposed to the light source duringthe aging process, experienced a net zero difference in its performance.Of note is PP02, which experienced a large instability soon after the lightsource was turned off. This instability occurred over less than a week andchanged by about 10% with respect to the initial value at 0 C. This mostlikely an outlier, with the rest of the PP having a more stable behaviourafter aging. The RMS of the PP performance was on average (2.19±0.26)%with respect the mean relative gain × quantum efficiency of each PP.56Chapter 8ConclusionThe R11283 Hamamatsu photopentode (PP) is a low gain, fine mesh photo-multiplier tube intended for use in the Belle II endcap calorimeter in conjunc-tion with pure CsI scintillation material. The design of the device optimizesperformance in a high magnetic field and large anode currents.The electronic noise was measured in two ways: using cosmic muons itwas estimated to be (77 ± 2) keV, and as a side effect of measuring theexcess noise factor it was found to be (32±1) keV, although with a differentCsI crystal. In the first case, it is thought that there may be some violatedassumptions when passing the PDF to the fitting program in ROOT. Inthe second case, a smaller crystal of higher quality was used to determinethe energy to DAQ output calibration, which is the source of the smallerequivalent noise energy. From the excess noise factor measurements, theelectronic noise of the preamp was found to be equivalent to 1730 ± 33electrons at the anode.Using a small pure CsI crystal and a few radioactive calibration sources,the variation with temperature was found to be (−1.65 ± 0.02) %/◦C, andit was discovered that the time until the dynodes became thermally equili-brated was on the order of 8 to 12 hours. The PP stability after thermalequilibrium was (0.28± 0.03)% for time scales on the order of a week. ThePP was also found to have a linear energy response. The slope of this re-sponse, proportional to the gain × quantum efficiency, is correlated withmany of the measurements Hamamatsu provides with their shipments. Asummary of the values Hamamatsu measured for the set of PP used in theseexperiments can be found in table 5.1.A relationship of interest to providing quality control parameters wouldbe to correlate the change in PP performance from the aging measurement(albeit with additional statistics) with a quantity that is measurable insmaller time scales. This quantity could be any of Hamamatu’s standardmeasurements that are currently in place, or something novel such as theexcess noise factor.The gain of the PP at an operating voltage of −1000 V was on average255 ± 11, in stark contrast to the gain of more conventional PMTs. The57Chapter 8. Conclusiongain of the PP was also found to be linear with the operating voltage, andhad an average rate of change of (0.335± 0.016) V −1.A test input on the preamp was used to calibrate the DAQ. With apulsed UV laser, the excess noise factor for the R11283 PP was found tobe (1.9 ± 0.1 ± 0.4) on average. The excess noise factor was found to varynon-linearly with the operating voltage, rising to 3.5± 0.1 at a gain of 25.The lifetime of the PP was measured by inducing a current of about2 µA though the anode by means of a UV light source. The product ofthe gain and the quantum efficiency was used to track the PP performanceand was seen to decrease to (93 ± 3)%, when averaged over 14 devices.Fifteen of the sixteen PP were aged for 48 days, with one of the aged PPsmeasuring the current only, and one PP remaining unaged as a control. Anintegrated current of 7C was passed through the anode. This is equivalentto about 70 years of expected standard Belle II operation. Most of theaging occurred within the first coulomb of charge, however there is someevidence that there are other factors at work in the aging of the PP. TheRMS, normalized by the mean of the PP performance relative to the firstmeasurement was (2.19± 0.26)% across the 14 PP.Large scale tests of 19 stage fine mesh PMTs for the ZEUS experimentshowed that the deviation in PMT performance increases as the aging pro-gresses [32]. A similar study for the E787 experiment at Brookhaven (BNL)showed that the gain of their model smoothly drops to 85% of the initialvalue after 430 C, but after 10000 hours of measurement they observed thatthe gain returned to 100% [33]. They also observed an increase in the scatterof the PMT gain. This, with the results from figures 7.13 & 7.14 suggeststhat a long term study of a greater number of PP is needed to properly char-acterize the aging process. This is especially needed if the aging is not onlydependent on the charge passing through the anode. It may be of interestto attempt to determine if there are alternate sources of aging.While the presented study has corrected for temperature and the DAQpedestal, there may be a real time delay to these measurements, as there waswith others. For instance, it was observed that there exists a delay betweenimmersing the PP in an environment and for the PP to reach thermal equi-librium with that environment. For the case of use in a thermally controlledarea, especially if the thermal conditions greatly differ from typical roomconditions, this delay would help determine the amount of dead time neededbetween servicing the PPs and being able to take reliable measurements. Asimilar delay time related to the aging may need to be characterized for theperiod prior to taking measurements. In case of delays on the order of anhour, one would need a new method of tracking the PP performance while58Chapter 8. Conclusionthe aging is in progress, in order to compare the performance during agingand between aging sessions.59Bibliography[1] C. Hearty, E. Ji, and P. Lu, “First Results with the Montreal Photopen-tode Preamp at TRIUMF.” Internal notes from ECL group meeting(http://kds.kek.jp/conferenceDisplay.py?confId=14102), Octo-ber 2013.[2] C. Hearty, “Updated Measurements on the Performance ofthe Montreal Photopentode Preamplifier.” Internal notes fromECL group meeting (http://kds.kek.jp/conferenceDisplay.py?confId=14434), December 2013.[3] E. Rutherford, “The Scattering of α and β Particles by Matter and theStructure of the Atom,” Philosophical Magazine., vol. 21, pp. 669–688,1911.[4] Belle, A.Abashian, et al., “The Belle Detector,” Nuclear Instrumentsand Methods in Physics Research A, vol. 479, pp. 117–232, 2002.[5] Belle, K.Miyabayashi, et al., “Physics Achievements from the Belle Ex-periment,” Prog. Theory. Exp. Phys., 2012. DOI: 10.1093/ptep/pts072.[6] Belle II, T.Abe, et al., “Belle II Technical Design Report,” 2010.arXiv:1011.0352.[7] BABAR, B. Aubert, et al., “The BABAR Detector,” Nuclear Instru-ments and Methods in Physics Research A, vol. 479, pp. 1–116, Febru-ary 2002.[8] Belle, K. Abe, et al., “Observation of Large CP Violation in the NeutralB Meson System,” Phys. Rev. Lett., vol. 87, p. 091802, Aug 2001.[9] BABAR, B. Aubert, et al., “Observation of CP Violation in the B0Meson System,” Phys. Rev. Lett., vol. 87, p. 091801, Aug 2001.[10] Nobelprize.org, “The Nobel Prize in Physics 2008.” http://www.nobelprize.org/nobel_prizes/physics/laureates/2008/.60Bibliography[11] G. Isidori, Y. Nir, and G. Perez, “Flavor Physics Constraints for PhysicsBeyond the Standard Model,” Annual Review of Nuclear and ParticleScience, vol. 60, no. 1, pp. 355–380, 2010.[12] A. Kuzmin, “Endcap Calorimeter for SuperBelle Based on Pure CsICrystals,” Nuclear Instruments and Methods in Physics Research Sec-tion A: Accelerators, Spectrometers, Detectors and Associated Equip-ment, vol. 623, no. 1, pp. 252 – 254, 2010. 1st International Conferenceon Technology and Instrumentation in Particle Physics.[13] K. Miyabayashi, “Belle Electromagnetic Calorimeter,” Nuclear In-struments and Methods in Physics Research Section A: Accelerators,Spectrometers, Detectors and Associated Equipment, vol. 494, no. 13,pp. 298 – 302, 2002. Proceedings of the 8th International Conferenceon Instrumentation for Colliding Beam Physics.[14] Saint-Gobain Crystals, “CsI(pure) Cesium Iodide Scintillation Mate-rial,” tech. rep., Saint-Gobain Ceramics & Plastics, Inc., 2007. AccessedAugust 3, 2015 at http://www.crystals.saint-gobain.com.[15] Saint-Gobain Crystals, “CsI(Tl), CsI(Na) Cesium Iodide ScintillationMaterial,” tech. rep., Saint-Gobain Ceramics & Plastics, Inc., 2007. Ac-cessed August 3, 2015 at http://www.crystals.saint-gobain.com.[16] C. Amsler, D. Grgler, W. Joffrain, D. Lindelf, M. Marchesotti,P. Niederberger, H. Pruys, C. Regenfus, P. Riedler, and A. Ro-tondi, “Temperature Dependence of Pure CsI: Scintillation Light Yieldand Decay Time,” Nuclear Instruments and Methods in Physics Re-search Section A: Accelerators, Spectrometers, Detectors and AssociatedEquipment, vol. 480, no. 23, pp. 494 – 500, 2002.[17] Belle II Collaboration, “Belle II Wireframe.” http://belle2.kek.jp/images/BelleII-outline.pdf. Image altered. Photo accessed May 11,2015.[18] J. Haba, “Letter of Intent for KEK Super B Factory, Part II: Detector,”April 2004. https://belle2.cc.kek.jp.[19] PANDA, A.Borisevich, et al., “PWO-II Scintillation Crystals for thePANDA Electromagnetic Calorimeter,” in IEEE Nuclear Science Sym-posium Conference Record, pp. 2698–2700, 2008.61Bibliography[20] B. Shwartz, “Belle Calorimeter Upgrade,” Nuclear Instruments andMethods in Physics Research Section A: Accelerators, Spectrometers,Detectors and Associated Equipment, vol. 598, no. 1, pp. 220 – 223,2009. Instrumentation for Collding Beam Physics Proceedings of the10th International Conference on Instrumentation for Colliding BeamPhysics.[21] Hamamatsu Photonics K.K., “Photomultiplier Tubes: Basics and Ap-plications,” 2007.[22] Hamamatsu Photonics K.K. Electron Tube Division, “PhotomultiplierTube R11283 Technical Data,” tech. rep., Hamamatsu Photonics, April2013.[23] J.-P. Martin, N. Starinski, and P. Taras, “Fast Charge-sensitive Pream-plifier for Pure CsI Crystals,” Nuclear Instruments and Methods inPhysics Research Section A: Accelerators, Spectrometers, Detectors andAssociated Equipment, vol. 778, pp. 120 – 125, 2015.[24] A. Hershenhorn, “CsI Cosmic Muon Deposit Energy.” Private commu-nication.[25] S. Ritt et al., “Maximum Integrated Data Acquisition System,” 1993-.https://midas.triumf.ca.[26] M. W. Davidson, “Tungsten-Halogen Incandescent Lamps.” AccessedAugust 13, 2015. http://zeiss-campus.magnet.fsu.edu/articles/lightsources/tungstenhalogen.html.[27] A. Kuzmin, “Endcap Calorimeter for SuperBelle based on Pure CsICrystals,” Nuclear Instruments and Methods in Physics Research Sec-tion A: Accelerators, Spectrometers, Detectors and Associated Equip-ment, vol. 623, no. 1, pp. 252 – 254, 2010. 1st International Conferenceon Technology and Instrumentation in Particle Physics.[28] A. Karar, Y. Musienko, and J. Vanel, “Characterization of AvalanchePhotodiodes for Calorimetry Applications,” Nuclear Instruments andMethods in Physics Research Section A: Accelerators, Spectrometers,Detectors and Associated Equipment, vol. 428, no. 23, pp. 413 – 431,1999.[29] Y. Jin, H. Aihara, O. Borshchev, D. Epifanov, S. Ponomarenko, andN. Surin, “Study of a Pure CsI Crystal Readout by APD for Belle II62BibliographyEnd Cap ECL Upgrade,” Nuclear Instruments and Methods in PhysicsResearch Section A: Accelerators, Spectrometers, Detectors and Asso-ciated Equipment, pp. –, 2015.[30] S. de Jong, “ECL backgrounds in the 11th campaign,” in 21st Belle 2General Meeting, 2015.[31] C. Hearty, “Initial Studies with SICCAS CsI Crystal and Preamp V3.”Internal notes from ECL group meeting, April 2014.[32] T. o. Ishii, “Automatic Test of Photomultiplier Tubes for the ZEUSForward and Rear Calorimeters,” Nuclear Instruments and Methods inPhysics Research A, vol. 320, pp. 449–459, Jan 1992.[33] T. K. Komatsubara et al., “Performance of Fine-mesh PhotomultiplierTubes Designed for an Undoped-CsI Endcap Photon Detector,” NuclearInstruments and Methods in Physics Research A, vol. 404, pp. 315–326,Feb. 1998.[34] Teledyne LeCroy, “Modular ADCs for Physics and Chemistry.” Ac-cessed August 12, 2015. http://teledynelecroy.com/lrs/dsheets/2249.htm.[35] Saint-Gobain Crystals, “Detector Assembly Materials,” tech. rep.,Saint-Gobain Ceramics & Plastics, Inc, 2005. http://www.crystals.saint-gobain.com/uploadedFiles/SG-Crystals/Documents/Organic%20Product%20Accessories%20Data%20Sheet.pdf.[36] Momentive Perfomance Materials, “TSE3032 Technical Data Sheet,”tech. rep., Momentive Perfomance Materials, 2015. Acessed 08/07/15.https://www.momentive.com/products/showtechnicaldatasheet.aspx?id=10404.[37] Hamamatsu Photonics K.K., “R5113-02 Photomultiplier Tube DataSheet.” Accessed Aug 10, 2015 at http://www.hamamatsu.com/jp/en/R5113-02.html.[38] R. B. Firestone, “The Berkeley Laboratory Isotopes Project’s Exploringthe Table of Isotopes,” May 2000. Accessed on August 13, 2015. http://ie.lbl.gov/education/isotopes.htm.[39] H. Ikeda et al., “A Detailed Test of the CsI(Tl) Calorimeter for BELLEwith Photon Beams of Energy Between 20 MeV and 5.4 GeV,” Nuclear63Instruments and Methods in Physics Research Section A: Accelerators,Spectrometers, Detectors and Associated Equipment, vol. 441, no. 3,pp. 401 – 426, 2000.64Appendix ACircuit DiagramsA.1 Electronic NoiseBelle II Electronic Noise Derek Fujimoto02/05/14PPPMTPMTHVDC VoltageOscilloscopeLeCroy 821ZQuad DiscriminatorLeCroy 365ALNIM Logic Ortec 474 Timing Filter AmpInInInOutOut OutInOutDark Box-3V+9VCoincidence 2Figure A.1: Circuit diagram for the electronic noise measurement in chapter3.65A.2. Short Term StabilityA.2 Short Term StabilityPreampPowerHV PowerR11283 PPNIM Logic UnitLRS 635ALPCADCLeCroy 2259B1000V-3V9VTiming Filter AmpOrtec 474Gain: 125Int/Diff: 50nsInverting OutputVetoOutInInOutDark BoxSignal InCAMAC Crate Model 1500InOutGate500ns Width195mV Threshold100ns WidthDsicrminatorLeCroy 821Z0DsicrminatorLeCroy 821ZAmplifierLeCroy 612AGain: 10DelayLineGate Generator LRS 222Out120us NIMStartFigure A.2: Circuit diagram for the short term stability measurement inchapter 4.66A.3. Excess Noise FactorA.3 Excess Noise FactorPreampPreamp PowerAttenuatorLeCroy A101DiscriminatorLeCroy 821ZPulse GeneratorBNC 8010AmplifierLeCroy 612APCADCLeCroy 2259B9V-3VTiming Filter AmpOrtec 474Gain: 125Int/Diff Time: 50nsInverting OutputGain: 10Gain: 0.3GateSignalCAMAC Crate Model 1500Figure A.3: Circuit diagram for the calibration pulse measurement in chapter6.67A.3.ExcessNoiseFactorPulserPowerGate GeneratorLRS 222PreampPowerHV PowerR11283 PPR5113 PMTNIM Logic UnitLRS 635ALPCLaser PowerPulsedLaserADCLeCroy 2259BADCLeCroy 2249Pulse GeneratorBNC 80101000V2200V-3V9VTiming Filter AmpOrtec 474Gain: 125Int/Diff: 50nsInverting OutputTTLGate GeneratorLRS 222NIMStart120us NIMStartVetoOutInInOutDark BoxSignal InSignal InCAMAC Crate Model 1500In OutGateGate500ns Out30mV Threshold100ns OutputDsicrminatorLeCroy 821Z0DsicrminatorLeCroy 821ZAmplifierLeCroy 612AGain: 10Figure A.4: Circuit diagram for the excess noise factor measurement in chapter 6.68A.4.AgingandLifetimeA.4 Aging and LifetimeLifetime MeasurementDerek FujimotoJuly 201516 PP Array HV Distributor HV SourceU de MontrealMotherboard(shapers)Delay Line (128ns)L2259B L2259BLogic Unit LRS635AL LRS2228(Provides LAM only)CAMAC CrateDark BoxIncubatorStartStopSignalSignalGate GateOutInDiscriminatorLeCroy 821ZGate GeneratorLRS222 VetoInOutInLeCroy AmpModel 612AKeithley PicoammeterMcGill LED FlasherBNC Pulse GeneratorModel 8010PCDiscriminatorLeCroy 623BLZ DiscriminatorLeCroy 623B Logic Fan-In/Fan-OutLeCroy 429ALogic Unit LRS635AL OutInOutOutOutInOutOutInOutFigure A.5: Circuit diagram for the aging and lifetime measurement in chapter 7.69Appendix BEquipment SpecificationsB.1 ElectronicsB.1.1 U. de Montre´al Preamp V3The purpose of the preamp is to produce a step function whose amplitudeis proportional to the integrated current passing through the anode over agiven time. This step function is sent to a shaper that produces a signalwhose amplitude is proportional to the amplitude of the preamp output.What is generally referred to in the main body of this work as “the preamp”is the printed circuit board housing both the voltage divider and the preampcircuit. Version 3 of this board can be seen in figure 2.3. The board has sixsoldered cables: two voltage inputs to power the preamp circuit, one highvoltage (HV) input for the voltage divider, a signal out cable, a calibrationtest pulse input, and a ground strap.The preamp power inputs were set to +9 V and -3 V and were connectedto an external DC power supply.The HV input goes directly to the high voltage divider. This circuit(fig. B.1) is isolated from the preamp circuit and distributes the correctamount of voltage to the photocathode, dynodes, and the anode of the PP.The calibration test pulse input leads to a (1±0.25) pF capacitor, whichallows for the injection of a known amount of charge into the preamp. TheFigure B.1: Universite´ de Montre´al voltage divider [23].70B.1. Electronicspreamp then processes this charge as if it were a real signal. This is usefulfor calibrating the readout electronics and for determining correspondingthe number of electrons from the shaper.The preamp ground strap was clamped to a larger ground strap, whichoften served as the ground for the dark box as well. This ground strap wasultimately connected to the building’s electrical ground, through the NIMcrate.The preamp and voltage divider are connected to the PP via five fly-ing leads, which are inserted through the board. The ends of the flyingleads were covered in electrical heat shrink to prevent sparking, which wasotherwise a significant source of noise.B.1.2 U. de Montre´al Preamp V4 and MotherboardsVersion 4 of the preamp (fig. B.2) did not have many functional changesfrom V3 [23]. The primary change was the inclusion of a mini-display portcable to carry the -3 V and +9 V voltages, the signal out, the calibrationpulse signal, and the ground to the preamp. The HV cable was updated toFigure B.2: Universite´ de Montre´al V4 preamp.a SHV cable, eliminating the need for an adapter. The mini-display portcables were connected to a motherboard (fig. B.3) which housed the shapersand distributed power to the preamps. The motherboard is powered by fourDC voltages: ±12 V and ±6 V via a DB9 connector, and has a separate71B.1. ElectronicsFigure B.3: Universite´ de Montre´al motherboard and shapers.input for the calibration pulse. Each motherboard contains housing for 8shapers, which are removable. The signal output of the shapers are lemoconnectors.There were a few issues when using the mini-display port cables. Thecables were single sided: only one end of the cable had the standard con-nector, whereas the other end had a custom port to connect to the preamp.The custom end of the cable was extremely fragile, and was easily damaged.Additionally, this end pulled out of the preamp socket quite easily, renderingany usage with an inaccessible preamp difficult. The other end of the cableworked extremely well. For future iterations of the preamp, it would bebest to keep the standard connector on both ends of the mini-display portcables. In this way, the cables can be exchanged with others, in the casewhere there is damage or a cable of a different length is needed. This wouldalso provide a more secure connection to the preamp.The primary issue with the motherboards was signal pickup from theDB9 power input. This was because the cables initially used were not coax-ial. The solution was to solder lemo cable connectors to the ends of theDB9 inputs, and to install lemo feedthroughs in the dark box. Since themotherboards were encased in the shielded dark box, this reduced pickupimmensely.72B.1. ElectronicsB.1.3 Modifed PreampWith the V4 preamps, U. de Montre´al also included a modifed preamp(fig. B.4) which allowed for the direct reading of current from the anodeof the PP. This was specifically designed for use with the Keithley 6485Picoammeter. For the most part the modified preamp performed well, withthe exception of a weak mechanical connection between the resistors and theprinted board. The modified preamp does not house the preamp electronicspresent in the other preamps and did not allow for the readout of signals inthe same manner.Figure B.4: Universite´ de Montre´al modified preamp for reading current.The modified preamp current output was measured as a function of thePP operating voltage, to convert the baseline measurements to the propervalues. This is seen in figure B.5 with a linear fit. The current was averagedover a period of 5 minutes.73B.1. ElectronicsOperating Voltage (-V)450 500 550 600 650 700 750Current (A)22.533.544.556−10×/ ndf 2χ 2275 / 3Prob 0intercept 09−7.389e±06 −1.746e−slope 11−1.394e±09 −8.823eFigure B.5: Current from the modified preamp, connected to the control PP,as a function of operating voltage, with linear fit.B.1.4 LeCroy L2249 and L2259BTwo LeCroy ADCs were used in conjunction with the CAMAC crate andthe MIDAS DAQ software: the L2259B peak sensing ADC to measure theshaper output on signals from the PP, and the L2249 integrating ADC tomeasure output from a standard PMT. The standard PMT used in this ex-periment was the Hamamatsu R5113-02. Despite their differences in func-tionality, the ADCs are very similar in design. Each is a 12-channel singleslot NIM unit with a test input and gate, and have many of the same designspecifications [34]. The L2259B accepts inputs from 0 V to −2 V with risetimes longer than 50 ns. The rise time was discovered to be an important re-quirement for in-going signals. If the gate start was within 50 ns of the peak,the ADC failed to find the proper amplitude for the signal. The L2259Balso has a 106 µs digitization time, which is why the circuits were designedto generate 120 µs of dead time between signals. Additionally, as seen infigure B.6, the ADC output was non-linear for small pulses. In contrast, itwas found that the L2249 ADC was linear over its full range.74B.2. OpticsInput Pulse Height (mV)500 1000 1500 2000 2500 3000Output Pulse Height (bin)050100150200250Calibration Pulse Height Test: MIDAS MeasurementsFigure B.6: The L2259B LeCroy ADC is nonlinear for input pulses scoringbelow bin 50. The measured pulse is the output of the calibration test pulsefrom the preamp. The x-axis is the calibration pulse input voltage.B.2 OpticsB.2.1 Optical GreaseThe optical grease used was the BC-630 Silicone Grease manufactured bySaint-Gobain. This model of grease has a 95% optical transmission forwavelengths between 280 nm and 700 nm. However, the transmission dropsto nearly zero below 270 nm. The grease has an index of refraction of 1.465and has low evaporation at room temperature [35]. Recall that the lightproduced by the scintillation of CsI peaks at 315 nm, making this opticalgrease a good candidate for the optical connection.It was observed that the grease formed a light mechanical connectionbetween the crystal and the PP. However, if a large amount of grease wasused, the light yield read out by the PP was diminished. Additionally, aftersome time the grease would react with the CsI and become slightly yellowed.The grease had to be renewed every few weeks to maintain a high qualityoptical connection.B.2.2 Silicone RubberTo glue to PP to the CsI crystals, a two part transparent silicone rub-ber was used. This rubber was produced by Momentive Performance Ma-terials, and has the product number TSE3032(A) and TSE3032(B). The75B.2. Opticsrubber cures when the two types are mixed. TSE3032 has a refractionindex of 1.406 and a nominal mixing ratio of 100:10 (A:B) by weight [36].Figure B.7: CsI crystal being glued tothe PP using TSE3032. The tape isto prevent the crystal sliding off of thePP while it was drying.The mixing procedure is as follows:• Weigh a given amount ofTSE3032(A).• Add 10% of TSE3032(B).• Mix carefully.• Place mixture in a vacuum un-til air bubbles have vanished.To glue the CsI to the PP, thePP was clamped such that the faceof the PP was upwards and level, asin figure B.7. The mixed TSE3032was added carefully to the centerof the PP face. The CsI was thenslowly pressed onto the glue, withone edge down first, such that theglue only made initial contact at onepoint with the crystal. The face ofthe CsI was then eased down untilthe whole face was pressed onto thePP. The contact was then inspected:if there were any air bubbles or ifthe silicone rubber did not cover thewhole face, the CsI was removed,both surfaces were wiped down, andthe process was restarted.Since the silicone rubber wasquite viscous prior to being cured,the CsI was lightly taped downalong the sides to hold it in place. As an additional safety measure, a“fence” of tape was applied around the edges of the PP to ensure that theCsI did not fall off the PP in the case the aforementioned tape did not hold.This setup was left to cure overnight.B.2.3 Standard High-Precision PMTThe standard, high-precision, PMT used was the Hamamatsu R5113-02 pho-tomultiplier tube [37]. It has similar characteristics as the R11283 photopen-76B.2. Opticstode of chapter 2. It is a head-on type, with a 51 mm diameter window madeof UV glass. It is sensitive to light with wavelength in the range of 185 nmto 650 nm, peaking at 420 nm. It has a bialkali photocathode, and 12 dyn-ode stages in a linear structure. The gain of the particular tube used, withthe output being attenuated by a factor of 0.3, was (3.99 ± 0.15) × 106, asdetailed in section 6.3.1.B.2.4 Pulsed UV LaserThe pulsed UV laser used in the excess noise experiment of chapter 6 was acustom device constructed by Pierre Amaudruz of the TRIUMF DAQ group.The output wavelength of the light was 405 nm which was transmitted alongthe length of a fibre optic into the dark box. The pulser was run by twopower sources, one of which set the intensity of the light emitted. The laserpulses were triggered by a TTL logic signal, which was sent from a NIMpulse generator.The fibre optic used was the HPSC25 from thorlab.com, and has a nu-merical aperture of 0.100 ± 0.015. The numerical aperture describes theangles over which the system can accept or emit light. It is given byNA = n sin(θ), (B.1)where NA is the numerical aperture, n is the index of refraction in themedium in which the system is contained, and θ is the angle of acceptanceor emittance. Using n = 1 for air, means that the light exiting the fibreoptic is emitted at an angle no greater than θ = 0.100 ± 0.0015 radians.Therefore the radius of the illuminated portion of the screen is one-tenththe screen to laser distance. The largest screen to laser separation used was60 cm, therefore since the reflective screen was wider than 12 cm, all of thelight was reflected off, or absorbed by the black floor of the dark box.77B.2. OpticsB.2.5 Diffuse Reflective ScreenFigure B.8: Diffuse reflective screenused in excess noise measurements ofchapter 6.Seen in figure B.8, a diffuse reflec-tive screen was used to provide uni-form light to the PP and the stan-dard PMT. This screen reflected thelaser light provided by the pulsedUV laser in appendix B.2.4. Therewere several materials tested includ-ing: printer paper (new and yel-lowed), sandpaper of three differ-ent grits and colours, and mylarcopy paper. The copy paper was asemi-transparent tracing paper usedto copy blueprints. One side ofthe paper provided diffuse reflec-tions while the other produced pri-marily specular reflections. The fi-nal screen design was four layers ofthe mylar copy paper taped aroundthe edges on a G10 board, with thediffuse side outwards. Since boththe G10 board and the copy paperwere slightly transparent, a layer offine grit sandpaper was used behindthe board to provide a neutral background. Tape was required around theedges since the mylar picked up a lot of dust and became dirty quickly. Carewas taken to ensure that the lower layers of mylar were clean before beingtaped. The whole setup was 18 cm × 28 cm and taped to an aluminumsupport structure so that it remained upright and perpendicular to the PP.78B.3. Miscellaneous EquipmentB.2.6 McGill LED ArrayFigure B.9: LED array designed toproduce light that emulates the emis-sion spectrum of CsI. Designed andproduced at McGill University.An LED array (fig. B.9) was de-signed and constructed by AmielKollek of McGill University, withlight output closely matching thewavelength of maximum emittanceof CsI. The array was composedof four UV LEDs emitting lightof 335 nm, 315 nm, 310 nm, and280 nm. These were set in parallelwith each other and in series witha variable resistor so that largervoltages may be used. The arraywas supposed to deliver an amountof light to the PP that was com-parable to a CsI signal, but at amuch higher rate in order to agethe PP. It was observed that threeof the four LEDs had experienceda catastrophic failure and were ir-reparably damaged. This proba-bly occurred during the search for asuitable pulse generator to run theLEDs, as the model used at McGill was not readily available at TRIUMF.With the remaining LED (335 nm), the array was run with a DC powersource with signals reaching the PP at a rate of about 1 MHz.The LEDs were produced by SETi (Sensor Electronic Technology, Inc.)model number UVTOP***TO39FW, where *** corresponds to the fourwavelengths stated above. They are a flat window type with a typicalFWHM of 10 nm. The typical forward voltage of the LED is 5.5 V, withoptical power of 600 µW.B.3 Miscellaneous EquipmentB.3.1 Dark BoxFor all of the measurements presented in this work, the experimental setupwas encased in a dark box (fig. B.10). The purpose of the dark box wasthreefold: to create a low light environment where the photomultiplier tubes79B.3. Miscellaneous Equipmentcould be operated safely, to shield from electromagnetic interference, and toprovide a low humidity environment for the CsI. For the first few trials ofthe electronic noise measurement, the dark box left over from Eddie Ji andChristopher Hearty was used. However due to foreseen space constraints,a larger box was soon purchased and light-proofed by Chelsea Dunning.The light proofing was done with the standard PMT and a optically opaquetape. The dark box itself was made of steel with interior dimensions ofabout 43.5 cm in height, 43 cm to 49.5 cm in width, and 119.5 cm in length.A layer of foam was taped around the open edge of the box, such that itwas compressed when the lid was shut, and the latches were taped over toensure light-tightness.Figure B.10: The dark box, prepared by Chelsea Dunning, that was used forall but the most preliminary measurements.The box was also grounded with a ground strap. The paint in a smallarea on the lid and body of the box was sanded off and bolts with wingnutswere installed to clamp the ground strap to both the body and the lid. Onthe inside, another ground strap was clamped to the body in the same man-ner, such that the box remained light-tight while allowing the electronicsinside to be grounded properly. The grounding of the box played a signif-icant role in reducing noise from pickup. Feedthroughs were installed forall necessary cables, except for the humidity and temperature probe, whichwas not readily detachable from the main device.B.3.2 DesiccantTo keep the humidity low, desiccant was used to absorb water from the air.Two types of desiccant were used: clay and a molecular sieve. As reported80B.3. Miscellaneous Equipmentby Chelsea Dunning, the molecular sieve was more effective at reducing thehumidity than the clay type. The molecular sieve also had the advantagethat it could be regenerated. This was performed by placing a tin of thedesiccant in an oven at 160 ◦C overnight, while flushing the oven with N2gas. Initially, a vacuum pump was used in place of the gas, but this idea wasretired after the vacuum was nearly damaged from water condensing out ofthe air while in the pump. The sieve used had a 4 A˚ pore size and was ofthe pellet type produced by Advanced Specialty Gas Equipment (ASGE).B.3.3 BINP CsI CrystalFor the electronic noise measurements presented in chapter 3, a pure CsIcrystal on loan from the Budker Institute of Nuclear Physics (BINP) wasused. As with the Belle crystals, this one was wrapped in a Gore-Tex filmand covered with aluminized mylar. It was a truncated pyramid with a crosssection of approximately 6 cm × 6 cm at the large end, 5.4 cm × 5.4 cm atthe narrow end, and about 30 cm in length. The PP was always connectedto the larger end of the crystal.Figure B.11: Temperature and hu-midity reader base for measurementswithin the dark box.As with all pure CsI, this crys-tal was kept in a low humidity envi-ronment to prevent damage to thecrystal surface.B.3.4 Temperatureand Humidity ProbeAs seen in figure B.11, the HH314ATemperature and Humidity Meterby Omega Engineering was used formeasurements in the dark box. Thisprobe came with software that couldbe used to monitor and record thetemperature and humidity at settime intervals via a computer. Theprecision of the temperature mea-surements was 0.1 ◦C.B.3.5 IncubatorThe incubator used was the 10-140 General Incubator produced by QuincyLabs, with interior dimensions 31×25×25 cm (fig. B.12). It is a heating-only81B.3. Miscellaneous Equipmenttype, therefore the set temperature needed to be larger than the ambienttemperature to ensure a stable environment.For the aging measurement, the standard window was replaced with aUVT acrylic window so that the UV LED light would age the PP properly.Additionally, the rear wall of the incubator was converted into a slidingpanel, so that the preamps and mini displayport cable connections could beaccessed. There was a small hole cut into the panel to let the SHV and minidisplayport cables out of the incubator.Figure B.12: Incubator used in experimental setup. Version shown has UVTwindow installed, and contains mechanical support structure for the agingmeasurements in chapter 7.82B.3. Miscellaneous EquipmentB.3.6 Calibration SourcesThree calibration sources were used: 137Cs 22Na and 207Bi . Below is atable of the energies of the common decays that produce gammas:Table B.1: Calibration source decay energies in keV [38]. Note that the511 keV decay from the 22Na source is from positron annihilation and is notdue to a 22Na gamma.137Cs 22Na 207Bi661.657 511.006 569.7021274.53 1063.6621770.23783Appendix CFunctionsC.1 NovosibirskAs described in reference [39], the Novosibirsk function is given byF (x;N, xp, σE , η) = N exp(− 12σ2oln2(1− x− xpσEη)− σ2o2), (C.1)whereσo =(2η)sinh−1(ηξ2)andξ = 2√ln 4 = 2.36.The variable parameters are as follows:• N : normalization factor• xp: peak location• σE : resolution, defined as the FWHM/ξ• η: asymmetry parameterThe Novosibirsk function is the result of the convolution of the energyspectrum of a Compton scattered photon and a log-normal distribution. Thetheoretical Compton energy spectrum is given byF (Eγ) = N((Eγ − Ec2)2+E2c4)(C.2)where Eγ is the energy of the Compton scattered photon and Ec is theCompton edge energy.In essence, the Novosibirsk function is a asymmetric semi-infinite Gaus-sian function.84C.2. Excess Noise FactorC.2 Excess Noise FactorStarting from the definition of the excess noise factor in chapter 6:(σcNc)2F =(σaNa)2, (6.1)it was noted that Nc, the number of electrons at the photocathode (photo-electrons), is proportional to the number of photons incident on the photo-cathode by a factor corresponding to the quantum efficiency. Therefore, thephotoelectrons are Poisson distributed and σ2c = Nc:FNc=σ2aN2a. (C.3)If the internal gain of the PP is given by M , then Na = M ·Nc. EquationC.3 does not take the electronic noise (σo) into account. The width ofthe distribution actually measured is given by σ2m = σ2o + σ2a, and the peaklocation is unchanged: Na = Nm. Substituting all of the above into equationC.3:σ2m = (FM2)Nc + σ2o .Where σm and σo are in units of number of electrons at the anode, and Ncis in units of number of photoelectrons. Furthermore, note that since σm,sis measured in number of electrons at the anode, then σm,s/M is in units ofphotoelectrons. Therefore,σ2mpe = F ·Nc + σ2ope , (6.2)where σmpe,ope are in units of number of photoelectrons. Since the excessnoise factor (F ), the internal PP gain (M), and the electronic noise (σo) areall constants, equation 6.2 is a simple linear relationship. For clarity, equa-tion 6.2 has been written in chapter 6 without the photoelectron indicators.C.3 Excess Noise Factor: Why the PMT WasNot UsefulIn chapter 6 it was stated that the original plan was to use a high precisionPMT to measure the number of incident photons and to use this informationto determine the excess noise factor of the PP. Since the standard PMT wasable to measure single photons, a direct measurement of the PMT gainwas possible, whereas the PP gain was not high enough to measure single85C.3. Excess Noise Factor: Why the PMT Was Not Usefulphotons. The issue with only using the PP was that it relies on using the gainas measured by Hamamatsu, which could not be checked with the existingsetup. What follows is the reasoning for why the high precision PMT wasnot useful, despite its significant advantage.Recall that the excess noise factor is given by:FNpe=σ2aN2a, (C.3)where Npe = Nc is the number of photoelectrons. Rearranging,F = Npe ·(σ2aN2a),and since σa and Na are known via direct measurement from the ADC, itfollows that there needs to be a measurement of Npe.C.3.1 Method 1: The Direct Measurement of NpeTo do this, note also the two following relationships:Npe = QENγ (C.4) Npe =NaM(C.5)where QE is the quantum efficiency, Nγ is the number of incident photons,Na is the number of electrons at the anode, and M is the gain of the pho-totube. With the setup described in chapter 6.2, there are two ways tomeasure Npe:• Eqn C.4: Nγ is measured and QE is applied (“through the front”).• Eqn C.5: Na is measured and M is applied (“through the back”).For the photopentode:• Both Nγ and QE are unknown, thus equation C.4 is not so useful.• Na is measurable, but M is known only from the Hamamatsu datasheet, making equation C.5 useful but undesirable.So a direct measurement of Npe using the PP would have to apply thegain as measured by Hamamatsu, with no way of checking this value.86C.3. Excess Noise Factor: Why the PMT Was Not UsefulC.3.2 Method 2: An Alternate Expression for NpeThe alternative to the direct measurement is to find an alternate expressionfor the Npe. Equation C.5 should not be used since any use of this relation-ship would reduce the alternate expression back to the direct measurementcase.It is useful to know that Nγ , although unknown, can be made the samefor both the PMT and the PP. This is because the setup allows for the twophototubes to be exchanged reliably, so if the signal from both phototubesdoes not change significantly upon exchange, it is reasonable to concludethat Nγ has not changed at either position. Using this with equation C.4:NpeN˜pe=QENγQ˜ENγ=QEQ˜E→ Npe = N˜peQEQ˜E(C.6)where the quantities with the tilde are measured with the standard PMT,and those without by the PP. Substituting equation C.6 into equation C.3givesF = N˜pe ·(QEQ˜E)·(σ2aN2a)F ·(Q˜EQE)= N˜pe ·(σ2aN2a)(C.7)where the excess noise factor is now known up to a constant multiplier.However, as was noted earlier, QE is unknown, and Q˜E is unknown as well.C.3.3 Conclusion: PMT Not UsefulThe two methods for solving for the excess noise factor are thenF =(NaM)·(σ2aN2a)or F ·(Q˜EQE)= N˜pe ·(σ2aN2a)where either the standard PMT is not useful (left) or the excess noise factorcan only be found to a multiplicative constant (right). It was decided tosimply make the direct measurement of Npe since this used only one PP,and had less room for error. As long as the Hamamatsu measurement ofthe gain of the phototube is accurate, method 1 should be viable. Thiswas why the standard PMT was included in all of the setup and proceduredesign, but the data from that PMT was not used to calculate the excessnoise factor of the PP.87C.4. Ratio of the Quantum EfficienciesC.4 Ratio of the Quantum EfficienciesFrom the prior analysis in chapter 6, the number of photoelectrons from eachphototube can be found. Seen in figure C.1 are the number of photoelectronsfor the same runs. It should be noted that the slope of this relationship isclose to one, the ratio of the number of photoelectons/ being equivalent tothe ratio of the quantum efficiencies (eqn. C.6).Npe (from PMT)60 80 100 120 140 160 180 200 220Npe (from PP)6080100120140160180200 / ndf 2χ 0.8084 / 2Prob   0.6675slope     0.04919± 0.9365 intercept  4.562±  4.55 Number of Phototelectrons at each PhototubeFigure C.1: Since the number of photoelectrons can be found for both thestandard PMT and the PP, their ratio should give the ratio of their quantumefficiencies. For this case, the ratio of the quantum efficiencies of the R5113-02 and the R11283 is given by the slope of the linear fit, 0.94± 0.05.88Appendix DAdditional ResultsD.1 Electronic Noise: Effect of Counter TiminghistFullEntries 2861Mean 0.1057±0.1858 RMS 0.07475±5.653 Underflow 0Overflow 1Time Offset (bin)-25 -20 -15 -10 -5 0 5 10 15Number of Events020406080100120(a) Bottom scintillator delayed.histFullEntries 619Mean 1.009RMS 31.81Underflow 0Overflow 1Integral 618-40 -20 0 20 40 60Number of Events Time Offset (bin)051015202530(b) Top scintillator delayed.Figure D.1: Time offsets for fits to the signals from the PP for differentdelays of the plastic scintillator counters used as the trigger. The distributionappears to be dependent on the order by which the trigger signals reach theoscilloscope.89D.2. Residuals for Energy LinearityD.2 Residuals for Energy LinearityEnergy (MeV)0.4 0.6 0.8 1 1.2 1.4 1.6 1.8(Peak - Fit) / Error -4-202468Standardized Residuals of Peak vs EnergyBi207Bi207Bi207Cs137Na22Na22Figure D.2: Standardized Residuals of figure 4.6. The squares are used tohighlight point locations only, and are not an indication of error.D.3 Gain as a Function of Operating VoltagePictured below are the gains measured by Hamamatsu as a function ofoperating voltage. In figure D.3, are the gains with linear fits, and in figureD.4 is a histogram of the slopes of these fits.Operating Voltage (-V)300 400 500 600 700 800 900 1000Internal Gain50100150200250Internal Gain for All 16 PPFigure D.3: Gain as a function of thePP operating voltage for all 16 PP.Note that the gains appear to con-verge around −300 V.h16Entries 15Mean 0.01631±0.3345 RMS 0.01154±0.06318 Underflow 0Overflow 0Slope of Gain vs Operating Voltage0 0.1 0.2 0.3 0.4 0.5 0.6Number of Entries0123456Slopes of Gain vs HV for All 16 PPFigure D.4: The distribution ofthe slopes from figure D.3. Theaverage rate of gain increase is(0.335± 0.016) V −1.90D.4. Calibration Pulse Results in Alternate UnitsD.4 Calibration Pulse Results in Alternate UnitsEffective Number of Photoelectrons100 150 200 250 300 350ADC-Pedestal4006008001000120014001600 / ndf 2χ  16.3 / 6Prob   0.01223intercept  2.827±17.18 − slope     0.01818± 4.689 Preamp to ADC Calibration for PP with Gain of 250.0Figure D.5: Calibration pulse test results of V3 preamp, as reported in sec-tion 6.3.4. The PP here was assumed to have a gain of 250, which is roughlythe average PP gain at −1000 V.Calibration Pulse Input Charge (C)4 6 8 10 12 1415−10×Preamp + Shaper + Amp Signal Amplitude (mV)400600800100012001400160018002000 / ndf 2χ 22.34 / 6Prob   0.00105intercept  4.322±  80.7 slope     5.59e+14± 1.244e+17 Preamp Ouput vs Input CalibrationFigure D.6: Calibration pulse test results of V3 preamp, as reported in sec-tion Justification of Novosibirsk Use in Excess Noise Factor Raw FitsEffective Number of Photoelectrons100 150 200 250 300 350Preamp + Shaper + Amp Signal Amplitude (mV)400600800100012001400160018002000 / ndf 2χ 22.34 / 6Prob   0.00105intercept  4.322±  80.7 slope     0.02239± 4.984 Preamp Ouput vs Input CalibrationFigure D.7: Calibration pulse test results of V3 preamp, as reported in sec-tion 6.3.4.D.5 Justification of Novosibirsk Use in ExcessNoise Factor Raw FitsIn chapter 6, the peak height spectrum was fitted with a Novosibirsk withthe claim that this was because the spectrum was asymmetric. Figures D.8and D.9 show that the Novosibirsk indeed fits the spectra better than aGaussian. This indicates that the spectrum is indeed asymmetric.run00000018Entries  83748Mean   0.2695±  404.7 RMS    0.1906±     78 Underflow       0Overflow        0 / ndf 2χ 976.4 / 590Prob   1.495e-21Constant  1.9± 433.8 Mean      0.3± 404.1 Sigma     0.19± 76.13 ADC200 300 400 500 600 700Number of Entries0100200300400500Gaussian Fit(a) Gaussian fit.run00000018Entries 83748Mean 0.2696±404.7 RMS 0.1906±78.01 Underflow 0Overflow 0/ ndf 2χ 555.6 / 589Prob 0.8349p0 1.9±433.2 p1 0.4±397 p2 0.20±76.68 p3 0.00316±-0.06504 ADC200 300 400 500 600 700 800Number of Entries0100200300400500Normalization Peak Width Tail Novosibirsk Fit(b) Novosibirsk fit.Figure D.8: Comparing the fits to the laser pulse spectra of chapter 6.92D.6. Screenshot of Calibration Pulse SignalADC100 200 300 400 500 600 700 800Data - Model-80-60-40-20020406080100Residuals for Gaussian Fit(a) Gaussian fit residuals.ADC100 200 300 400 500 600 700 800Data - Model-60-40-20020406080Residuals For Novosibirsk Fit(b) Novosibirsk fit residuals.Figure D.9: Residuals of the fits in figure D.8. It is seen that the Novosibirskfunction fits the data much better than the Gaussian.D.6 Screenshot of Calibration Pulse SignalFigure D.10: Screenshot of the average of 512 calibration pulse signals usingthe Tektronix oscilloscope. This signal is after the amplification and theshaping. The upper line is the signal from the preamp, while the lower lineis the gate used for the ADC trigger.93D.7. Details on the Aging MeasurementD.7 Details on the Aging MeasurementTime Elapsed (days)0 10 20 30 40 50Current (A)00.511.522.536−10×Figure D.11: Current passing through the modified preamp as a function oftime. The spikes occur when the LED is turned off to take a 207Bi spectrum.It was observed that the current oscillated over the course of the day. Thiswas most likely due to variations in temperature affecting the LED array,which was exterior to the incubator.data_L2249_15_1Entries  923022Mean   0.1701±  253.3 RMS    0.1203±  163.4 Underflow       0Overflow        0 / ndf 2χ  1209 / 1062Prob   0.001051novo1_norm  11.0±  2045 novo1_peak  0.3± 230.3 novo1_width  0.33± 51.34 novo1_tail  0.0083±0.1778 − novo2_norm  4.1±   510 novo2_peak  0.6± 537.1 novo2_width  0.66± 69.35 novo2_tail  0.00951± 0.05359 novo3_norm  0.67± 17.33 novo3_peak  4.7± 969.2 novo3_width  7.43± 97.08 novo3_tail  0.0621± 0.3008 expOffset  0.01±  8.57 expScale  0.000060±0.006526 − ADC0 200 400 600 800 1000 1200 1400 1600 1800 2000Number of Entries010002000300040005000Three Peak Fit for PP01(a) Full range.data_L2249_15_1Entries  923022Mean    1.074±  854.6 RMS    0.7591±  126.9 Underflow       0Overflow        0 / ndf 2χ   1209 / 1062Prob   0.001051novo1_norm  11.0±  2045 novo1_peak  0.3± 230.3 novo1_width  0.33± 51.34 novo1_tail  0.0083±0.1778 − novo2_norm  4.1±   510 novo2_peak  0.6± 537.1 novo2_width  0.66± 69.35 novo2_tail  0.00951± 0.05359 novo3_norm  0.67± 17.33 novo3_peak  4.7± 969.2 novo3_width  7.43± 97.08 novo3_tail  0.0621± 0.3008 expOffset  0.01±  8.57 expScale  0.000060±0.006526 − ADC700 800 900 1000 1100 1200Number of Entries020406080100Three Peak Fit for PP01(b) Zoomed on 1.7 MeV peak.Figure D.12: Fit to all three energy peaks of the 207Bi spectrum with the sumof three Novosibirsk functions and an exponential.To group like PP together, the slopes of a linear fit to the PP performance(figs 7.10 & 7.11) before and after a given charge were compared, as seen infigure D.13. Table D.1 describes the criteria for sorting. Note that under94D.7. Details on the Aging Measurementthis scheme, the control PP was labelled as a phototube that did not age atall.Table D.1: Categorization outline for the PP aging, where Sb and Sa are theslopes before and after 1C respectively. The values were chosen arbitrarilyto match a visual analysis of the progression of the PP performance. Forthe results of the categorization, refer to figure D.13.Colour Attributed Significance RequirementGreen Aged (burned), then stopped Sb/Sa > 4 and Sa > −0.5Blue Aged(burned), then slowed Sb/Sa > 4 and Sa < −0.5Black Aged at a continuous rate Sb/Sa < 4 and Sa < −0.5Red Didn’t age at all Sb > 0 or Sa > 0None Control PPCharge Through Anode (C)0 1 2 3 4 5 6 7 8Relative Gain X QE0.880.90.920.940.960.9811.021.04PP01Charge Through Anode (C)0 1 2 3 4 5 6 7 8Relative Gain X QE0.880.90.920.940.960.9811.021.04PP02Charge Through Anode (C)0 1 2 3 4 5 6Relative Gain X QE0.880.90.920.940.960.9811.021.04PP03Charge Through Anode (C)0 1 2 3 4 5 6Relative Gain X QE0.880.90.920.940.960.9811.021.04PP04Charge Through Anode (C)0 1 2 3 4 5 6 7 8Relative GainX QE0.880.90.920.940.960.9811.021.04PP05Charge Through Anode (C)0 1 2 3 4 5 6 7 8Relative GainX QE0.880.90.920.940.960.9811.021.04PP06Charge Through Anode (C)0 1 2 3 4 5 6 7 8Relative GainX QE0.880.90.920.940.960.9811.021.04PP07Charge Through Anode (C)0 1 2 3 4 5 6Relative GainX QE0.880.90.920.940.960.9811.021.04PP08Charge Through Anode (C)0 1 2 3 4 5 6 7Relative Gain X QE0.880.90.920.940.960.9811.021.04PP09Charge Through Anode (C)0 2 4 6 8 10Relative Gain X QE0.880.90.920.940.960.9811.021.04PP11Charge Through Anode (C)0 1 2 3 4 5 6 7 8Relative Gain X QE0.880.90.920.940.960.9811.021.04PP12Charge Through Anode (C)0 1 2 3 4 5 6Relative Gain X QE0.880.90.920.940.960.9811.021.04PP13Charge Through Anode (C)0 1 2 3 4 5 6 7Relative Gain X QE0.880.90.920.940.960.9811.021.04PP14Charge Through Anode (C)0 1 2 3 4 5 6 7Relative Gain X QE0.880.90.920.940.960.9811.021.04PP15Charge Through Anode (C)0 1 2 3 4 5 6 7 8Relative Gain X QE0.880.90.920.940.960.9811.021.04PP16Figure D.13: Relative PP performance with categorization, as a function ofcharge. The PP have been divided into four categories: aged for a period,then stopped (solid); aged, but didn’t stop (− − −); aged at a constant ratethroughout (· · ·); and no aging at all (− · −).95


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items