Fortgeschrittenenpraktikum: Light Sensors for γ-ray Astronomy

Physik Department - Technische Universita t Mu nchen Max-Planck-Institut fu r Physik Fortgeschrittenenpraktikum: Light Sensors for γ-ray Astronomy V 1.0 Christian Fruck, Priyadarshini Bangale cfruck@ph.tum.de, priya@mpp.mpg.de

Abstract In nearly all fields of astro-particle physics, the detection of the primary particles is based on the detection of light created in secondary processes. Such processes include: fluorescence, scintillation and the production of Cherenkov light. Light sensors used in astroparticle physics should offer a high Photon Detection Efficiency (PDE), ultra-fast time response and good charge resolution as well as low dark-noise rates. The most popular such devices are Photomultiplier tubes (PMTs), Hybrid Photo Diodes (HPDs) and the recently developed Silicon photomultipliers (SiPMs). This lab exercise is intended to give a basic knowledge about these detector types, introduce to standard techniques and teach some basic skills for measuring their properties. Title picture: PMTs of the MAGIC I camera behind their Winston cones (picture courtesy Daisuke Nakajima)

Contents 1 Introduction 4 1.1 Astro-particle background............................ 4 1.2 Detector types.................................. 5 1.2.1 Photomultiplier Tubes (PMT)..................... 5 1.2.2 Hybrid Photo Diodes (HPD)...................... 7 1.2.3 SIPM................................... 9 2 Experiments 13 2.1 QE Measurements................................ 13 2.2 Measuring Gain.................................. 14 2.3 Cross Talk in SiPMs............................... 14 2.4 After-pulses.................................... 15 3 Evaluation 17 3.1 QE Measurements................................ 17 3.2 Measuring Gain.................................. 17 3.3 Cross Talk in SiPMs............................... 17 3.4 After-pulses.................................... 17 3

1 Introduction 1.1 Astro-particle physics background Astro-particle physics connects the methods and the research goals of two otherwise more distinct fields of research: astronomy/astrophysics and particle physics. This way, one can profit from the synergy in both directions. The theoretical understanding of astronomical objects can be improved by adding observational information from new channels and a wider spectral range. Particle physics on the other hand can profit from the extreme environments that are present in such objects. This allows to study processes that can neither be found nor - by using known technology - be generated in any lab on our planet. Examples range from single exotic stellar objects over highly active galaxies until the existence and properties of dark matter in general. All charged particles have the disadvantage that their origin cannot easily be reconstructed. This is due to their deflection by randomly distributed magnetic fields that are present everywhere within our galaxy and even beyond. The only non charged particles available are photons and neutrinos. Neutrinos are not that easy to detect and require huge and very sensitive detectors. Photons on the other hand are easy to detect and exploited to the maximum by classical astronomy in the optical, infrared and radio regime. Today also space-borne X-ray instruments are quite well established. All these instruments have one thing in common, which is that they can use optics to collimate electromagnetic waves. At even higher energies, namely in the γ-ray and very high energy γ-ray regime one has to use different techniques. It is possible to detect them by using methods from particle physics, namely constructing huge detectors with trackers and calorimeters to determine origin and energy of those particles. However, this is only possible outside our atmosphere and up to certain energies ( 100 GeV). Beyond this limit, the required size of such a space-borne instrument is just too large allow its realization because the fluxes are very low and the particle showers are too big to confine. Figure 1.1 gives an overview of the whole electromagnetic spectrum that can be used for astronomy. This is when ground based techniques are cutting in again. In ground based γ-ray astronomy one uses the fact that primary high energy γ-particles ( 1 GeV) create extended air showers (EAS) in the atmosphere (see Fig. 1.2 for an illustration of the processes involved). Here it is possible to either directly detect the secondaries in water tanks for example or to use fluorescence or Cherenkov light to record images of air-showers. Cherenkov light is always generated when charged particles are moving faster than the speed of light c/n in an optical medium. It is emmited under an angle θ. cos(θ) = 1 nβ (1.1) The same detection techniques are also used for charged cosmic rays, like for example in the AUGER 1 experiment and also in underground experiments serching for neutrinos. 1 www.auger.de 4

1.2. DETECTOR TYPES CHAPTER 1. INTRODUCTION Figure 1.1: The electromagnetic spectrum with the optically transparent windows, suitable for direct ground based observations. The solid blue line indicates the height, where the incident power has become half due to atmospheric attenuation. Image taken from Wagner (2006). Examples for ground based γ-ray telescopes are MILLAGRO 2, HESS 3, VERITAS 4 and MAGIC 5 (see Fig. 1.3 for a picture of the MAGIC telescopes). The last three of them are using the Imaging Air-shower Cherenkov Telescope (IACT) technique, where an image of the air-shower is mapped to a very fast and sensitive camera by a large optical mirrortelescope. For such cameras or as sensors for fluorescence or scintillation light, detectors with sensitivity for very low light levels and fast response time are required. This lab exercise shall give some background knowledge and introduce some standard techniques, how to measure their properties and compare such devices. 1.2 Different photo-detector types All light sensors available in this lab exercise are intended for use in IACTs. However, the basic working principles of different detector types stays the same even if the requirements on some properties changes from field to field. 1.2.1 Photomultiplier Tubes (PMT) Photomultiplier Tubes or PMTs have a long tradition in applications, where fast light signals have to be converted into electrical pulses. The first PMTs were invented in the 1930s and allowed the development of the first light sensitive TV cameras (Kubetsky 2 www.lanl.gov/milagro 3 www.mpi-hd.mpg.de/hfm/hess 4 veritas.sao.arizona.edu 5 magic.mppmu.mpg.de 5

1.2. DETECTOR TYPES CHAPTER 1. INTRODUCTION Primary Cosmic Ray (p,, Fe...) Atmospheric Nucleus e+ eem Shower Nucleons, K, etc. e+ e+ Atmospheric Nucleus ee+ e- EM Shower e- Nucleons, K, etc. e+ e+ e+ e e+ e- EM Shower Figure 1.2: Illustration of the secondary processes leading to electromagnetic and hadronic airshowers (Wagner 2006) Figure 1.3: The two MAGIC telescopes located on the Canary island La Palma observe the night sky in the spectral range of very high energy γ-radiation. They detect the faint Cherenkov light flashes produced by air-showers using cameras equipped with over 1000 photomultipliers each (picture courtesy Robert Wagner). 6

1.2. DETECTOR TYPES CHAPTER 1. INTRODUCTION (1937), Zworykin et al. (1936), Iams & Salzberg (1935)). They are one of the few examples of vacuum tube devices still used today. The working principle of a PMT is described in Fig. 1.4. Photons hitting the photocathode of a PMT can create a free electron inside the tube. The number of released electrons over impinging photons is called quantum efficiency or QE. QE = N ph.e. N phot. (1.2) For the multiplication process, a dynode system (metal plates coated with low-workfunction 6 materials) is supplied with negative high voltage (typically -1000V), as well as the photo-cathode. A voltage divider is used to distribute the voltage in a cascading way from the photo-cathode to the anode. A special field configuration is used to accelerate and focus the electrons onto the first dynode (see Fig. 1.5). The energy that the e gained being accelerated by the electric filed is released in the impact and kicks out a certain number of new electrons, which are then accelerated to the next dynode and so on. Typically, the dynodes provide a charge multiplication of a factor 4-6. To improve charge resolution, for the first dynode one can use a higher voltage and reach charge multiplication of up to 15. In the end one can measure a substantial electric pulse at the anode. The overall efficiency for detecting a photon in such a device (photon detection efficiency - P DE) depends on the quantum efficiency QE(λ) and the ph.e. collection efficiency α(u). P DE(λ, U) = QE(λ) α(u) (1.3) Typical pulse lengths are of a few ns. The total charge gain η depends on the applied voltage and number of dynodes N and their individual gains η i (if different). η = N η i (1.4) i Typical values of total charge gain range from 3 10 4 to 1 10 8. Also nowadays as there are other options like for example Silicon photomultipliers available, classic PMTs will not at all loose their importance in astro-particle physics experiments. The still are the detector of choice for covering large active areas and low dark rates. 1.2.2 Hybrid Photo Diodes (HPD) So-called hybrid photo diodes or HPDs are a relatively new detector type with unique properties. The best models (using a GaAsP photocathode) offer a peak QE of around 50% and are capable of detecting single and multiple photon events with good resolution. The difference to a normal PMT is that the whole charge amplification happens in two stages only. HPDs combine a vacuum tube with single stage HV acceleration track (about -8kV) and a Silicon avalanche diode that serves as second stage (see Fig. 1.7 for explanation). The gain from the first stage (also called electron bombardment gain) is on the order of 1 10 3. The second stage provides a gain (avalanche gain) of the order 1 10 2. That gives 6 The term work-funktion is describing the energy that is needed to remove an electron from a material to vacuum 7

1.2. DETECTOR TYPES CHAPTER 1. INTRODUCTION Figure 1.4: Working principle of a Photomultiplier Tube: When a photon hits the photo-cathode, it can produce a free electron with a probability called quantum efficiency (QE). A voltage of -1000V is applied to the photo-cathode and with a voltage divider distributed in a cascading way over the dynode system (5-7 dynodes typically). The e is accelerated and focused onto the first dynode, where its kinetic energy releases a certain number of secondary electrons that are then accelerated onto the next dynode and so on. In the end, the charge collected on the anode side is big enough to create a measurable electrical pulse (Drawing from Hamamatsu Photonics KK Editorial Committee (2006)). Figure 1.5: Typical field (equipotentials) configuration for focusing electrons in a PMT to maximize the collection efficiency (drawing from Burle Industries (1989)). 8

1.2. DETECTOR TYPES CHAPTER 1. INTRODUCTION 60 50 40 QE [%] 30 20 10 0 200 300 400 500 600 700 800 wavelength [nm] Figure 1.6: Example for a QE curve of the HPD model that was developed for MAGIC and is also used in the detector of the MAGIC LIDAR (adapted from Fruck (2010)). a total gain of 10 5. The bombardment gain η b (U) can be estimated from the band gap in Silicon E g (3.6eV). But it has to be taken into account that part of the bombardment energy E lost is lost (on the order of 2.5keV) in the inactive outer layer of the avalanche diode. After all, the bombardment gain of an HPD can be written like that: η b (U) (e U) E lost E g (1.5) Figure 1.8 shows the HPD model R9792U-40 from Hamamatsu that was produced as prototype that could replace the PMTs in the MAGIC cameras. This model has a peak QE 55% at 532nm and is therefore also used in the MAGIC LIDAR (LIght Detection And Ranging) system, where it provides a unique sensitivity in combination with a very weak laser (5µJ at 1kHz) Fig. 1.9. 1.2.3 Silicon Photo Multipliers (SiPM) In the nineties a novel semiconductor photon detector concept (e.g Bondarenko et al. (2000), Golovin & Saveliev (2004), Buzhan et al. (2003)) was developed, which is a promising candidate for the photon detector in astroparticle and high energy physics experiments ((Otte 2007), (Otte et al. 2004)). Currently, the detector is in advanced phase. Figure 1.10 shows a MEPhI and Pulsar enterprise prototype SiPM that was produced in Moscow in 2003. The device has an active area of (1 1) mm 2 inside which 576 avalanche photo diodes (cells) implanted on a silicon chip. Each cell operates in the limited Geiger mode, i.e. each cell is biased via an integrated resistor to a voltage slightly above breakdown (as shown in Figure 2.1. In this mode an electrical breakdown of the reversely biased junction can be triggered by a single photon. The breakdown results in a well measurable output signal, which is independent of the number of photons that have been absorbed in the cell. All cells of a SiPM are connected to a common anode bus. The signals from all cells are added up on the bus and thus the output signal of a SiPM is the summed up signal of all coinciding in time fired cells. The 9

1.2. DETECTOR TYPES CHAPTER 1. INTRODUCTION entrance window focusing e - GaAsP photocathode -8kV +400V out avalanche diode Figure 1.7: Working principle of a hybrid photo detector (HPD): With the probability of the QE, an incident photon releases an electron from the photo-cathode, to which high voltage of (minus) several kv is applied. The e is accelerated and focused onto an avalanche diode, where it creates electron-hole pairs loosing its kinetic energy (bombardement gain). these electron hole pairs are then amplified by the avalanche diode (avalanche gain) (drawing from Fruck (2010)). Figure 1.8: Hamamatsu R9792U-40 MHP0128HPD. This detector was designed as a possible replacement for the PMTs of the MAGIC telescopes, therefore the hexagonal shape. It is used as a light detector in the MAGIC LIDAR system. (image from Fruck (2010)). 10

1.2. DETECTOR TYPES CHAPTER 1. INTRODUCTION Figure 1.9: The MAGIC LIDAR system, which is equipped with a HPD, and the MAGIC II telescope in the background. HPDs are very good detectors for low power elastic LIDAR, because they provide a very low dark rate and good singel photon resolution (picture by Robert Wagner). Figure 1.10: A prototype SiPM comprising 576 cells. Picture from Otte (2007). 11

1.2. DETECTOR TYPES CHAPTER 1. INTRODUCTION n V p n V p hole electron Proportional mode Geiger mode Figure 1.11: Multiplication process in an avalanche diode that is reverse-biased above breakdown voltage multiplication by electrons and holes (Geiger mode). In Geiger mode the avalanche diverges, whereas in proportional mode the avalanche stops automatically as the avalanche propagates only in one direction. (credit: Otte (2007)). SiPM has two terminals, one for the bias voltage and another one for the output signal. Care has to be taken that the output terminal is DC coupled to the ground. In comparison with PMTs, SiPMs are comparably fast, operate at low voltage, and insensitive to magnetic fields. In general, they have good photon detection efficiency (PDE), which eventually in future could lower the energy threshold of the IACTs, and they can be operated at high background illumination, hence can increase the instrument duty cycle. There are few drawbacks also, such as, only small size available, relatively high dark count rate, optical cross talk, afterpulses, their performance depends on the bias voltage and temperature, and PDE is limited by the geometrical factor. In SiPMs, the hot carrier gives rise to an effect called optical crosstalk which appears when the photons created in an avalanche can propagate mostly unhampered within the device and may be absorbed in the sensitive volume of a different cell, thus triggering an additional breakdown ((Otte 2007)). The PDE is given as a function of several factors: P DE(λ) = Geom eff Geiger eff (λ) T ransmit eff (λ) QE intrinsic (λ) (1.6) Where Geom eff is a ratio of the single cell light sensitive area to its total area, Geiger eff (λ) is a probability for starting a Geiger avalanche and it depends on overvoltage and temperature, T ransmit eff (λ) is wavelength-dependent transport of impinging photons into the sensitive volume, and QE intrinsic (λ) is an intrinsic quantum efficiency (S. Gentile & Meddi 2010). 12

2 The lab experiments What follows is a selection of experiments available to students in our labs. It is possible that one of the setups is not available because it is in use. Otherwise you can choose three of the following tasks: 2.1 Measuring Quantum Efficiency (QE) In this experiment you will use a measurement setup, consisting of a tunable light source and a calibrated reference detector to measure the quantum efficiency of a PMT and a HPD. The light source is a combination of a halogen lamp (visible and IR) and a deuterium lamp (blue and UV) that is attached to an optical grating and an order filter. The reference detector is a PIN photo-diode with known QE in all wavelengths, available in a spreadsheet file. The measurements are automatized with a LabVIEW program. Filters are selected automatically and the currents are read with a picoamperemeter. The experimental procedure is as follows: 1. Always switch on the light sources first! They need some time to stabilize. In the meanwhile familiarize yourself with the setup and the LabVIEW program. Your supervisor will help you and answer any questions. 2. Take a reference measurement: Install the PIN photo-diode into the setup. Close the box to be absolutely light-tight (You can also use the big peace of black cloth to improve the light-tightness). Then run the measurement program (Be careful with the output files - always remember your filenames. You will need them for data analysis later.). 3. Install the PMT in the measurement setup (Warning: NEVER switch on HV before you finally closed the box). The PMT is connected in a way that all dynodes are connected to together to the readout cable. The effect is that all charge that is generated on the photo-cathode is collected by the dynode system and read out by the pcicoamperemeter. No charge multiplication should take place. The HV you have to adjust to a value that assures maximum collection efficiency. When everything is installed in the box and the box is closed, slowly increase the HV until the current saturates. 4. Run the measurement program for the PMT. 5. Now install the HPD in the QE box (Again: Do not switch on HV before everything is inside the box and the box is closed.). For the HPD, again, you do not use the avalanche diode, but directly shortcut anode and cathode and connect them to the readout. Again you apply HV and let the current saturate for maximum collection efficiency. 6. Now you perform the last QE measurement run with the HPD. 13

2.2. MEASURING GAIN CHAPTER 2. EXPERIMENTS 2.2 Measuring gain and charge resolution In this experiment you will operate different light sensors in their nominal charge multiplication mode. A light source with very short pulses 1ns will be used. You will use an oscilloscope to record a charge histogram by integrating over many signal waveforms. The oscilloscope will be triggered with the frequency of the light source. From the recorded histograms you can derive the gain of the device and measure the charge resolution. 1. The PMT is installed in a dark box, together with pulsed light source of your choice (fast LED or LASER). The light source is pointed to a piece of diffusely reflecting material, in order to get a smaller light flux on and a more homogeneous illumination on the PMT. The typical HV value for the PMT model is applied after the box is sealed totally light-tight (Again, be very careful with the HV source. Only switch it on after you sealed the box and double checked the cabling. That is for your own safety as well as for the safety of the equipped). 2. The PMT signal output is connected to a charge sensitive amplifier (This one needs ± 15V on the power supply). The output of the amplifier goes to the oscilloscope, as well as the trigger output of the pulse generator for the light source. 3. Adjust the light intensity to a value that gives one photon (a few) per pulse on average for the gain (charge resolution) measurement. 4. Now the signal integration mode of the oscilloscope is selected (your supervisor will help you) and a large number of waveforms are analyzed and summed up. You should should get a histogram with a pedestal peak ( zero charge from trigger events with no photo-electron generated) and one or more peaks originating from one or more photo-electrons. The distance between the peaks gives you the charge resolution. 5. perform this measurement for at least 4 different HV values. 2.3 Single photoelectron and Cross-talk measurement for SiPMs The setup for SiPM single photoelectron and cross talk measurement constist of, two variable power supplies, an amplifier, a light source which will be either Laser (405/440 nm) or LED (500/598 nm), a picoammeter, and an oscilloscope. The measuremnts are automatized with a LabView program. For data analysis a ROOT macro is avilable. The experimental procedure is as follows: 1. Switch on oscilloscope first as it takes time to start. Meanwhile, check the setup inside the box including proper connections to SiPM circuit and to external amplifier. 2. Before placing SiPM in the circuit, always check the polarity as you need SiPM to work in reverse bias. Prior polarity check is very important, as wrong polarity can damage the SiPM. 3. Close the box properly. To make sure it is light tight, you can use additional black cloth also. 14

2.4. AFTER-PULSES CHAPTER 2. EXPERIMENTS Figure 2.1: Basic setup for single photoelectron measurement and crosstalk calculation for SiPMs (2.3) and measuring after-pulses of PMTs(2.4) (Credit: Daniel Mazin). 4. Get familiar with oscilloscope settings, your supervisor will help you in it initially. 5. Switch on both power supplies. Apply ±15 V for amplifier and voltage for SiPM will vary according to the type. 6. Apply trigger and set the intensity level for Laser/LED. 7. Check if you could detect the signal from SiPM. 8. Run the measurement and take data using the LabView program or directly in the oscilloscope. 9. Repeat the procedure for different intensity levels of Laser/LED. 2.4 Measuring after-pulses of PMTs A problem, that all impact multiplication light detectors have in common is that of afterpulsing due to ion impact on the photo-cathode. The process behind is the following: An electron that was created by normal photon impact on the photo-cathode hits an atom of the rest gas or some other impurity of the vacuum. This can either happen on the way to the first dynode or more likely on the first dynode on some adhesively bound atom/molecule. This can create a partially ionized particle that is accelerated backwards towards the cathode due to its positive charge. A lot of effort is made in terms of field configuration (see Fig.: 1.5) to prevent such ions to make their way to the cathode but still it can not be excluded completely. Finally, the impact of such a particle releases a whole bunch of electrons that are amplified by the dynode system and create big after-pulses. The time delay, in respect to the initial light pulse, of such after-pulses is characteristic, depending on their charge and mass. 15

2.4. AFTER-PULSES CHAPTER 2. EXPERIMENTS To get a better feeling for this effect, you will use a pulsed light source and pile up the obtained signal of many pulses in order to identify different populations of after-pulses. The measurement consists of the following steps: 1. The PMT is installed in a dark box, together with pulsed light source of your choice (fast LED or LASER). The light source is pointed to a piece of diffusely reflecting material, in order to get a smaller light flux on and a more homogeneous illumination on the PMT. The typical HV value for the PMT model is applied after the box is sealed totally light-tight (Again, be very careful with the HV source. Only switch it on after you sealed the box and double checked the cabling. That is for your own safety as well as for the safety of the equipped). 2. The PMT signal output is connected to a charge sensitive amplifier (This one needs ± 15V on the power supply). The output of the amplifier goes to the oscilloscope, as well as the trigger output of the pulse generator for the light source. 3. Adjust the light intensity to a value that gives a few photons per pulse on average. Monitor the oscilloscope for a while. Maybe you can already spot some after-pulse events. 4. Now the pile-up mode of the oscilloscope is selected (your supervisor will help you) and a large number of waveforms is summed up. You should be able to see different populations of after-pulses if you selected the settings correctly. 16

3 Evaluation of the results 3.1 Measuring Quantum Efficiency (QE) Produce a plot of QE over wavelength for every detector that you have tested. 3.2 Measuring gain and charge resolution Produce a plot of gain over HV for the PMT(s)/HPD that you characterized. Compare the charge resolution of both detector types. Can you explain the differences? 3.3 Single photoelectron measurement and Cross-talk calculation for SiPMs i) Produce a histogram showing SiPM single photoelectron(s). Locate the pedestial and photoelectrons seperately. Compare the results at different voltages and intensities. Do you expect any difference? ii) Produce a plot for Cross talk vs voltage at different votage and intensity settings. Compare and explain the differences. 3.4 Measuring after-pulses of PMTs At what time delays could you observe after-pulses? Is it possible to assign certain elements/molecules to the populations? 17

Bibliography Bondarenko, G., Buzhan, P., Dolgoshein, B., et al. 2000, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 442, 187 Buzhan, P., Dolgoshein, B., Filatov, L., et al. 2003, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 504, 48 Fruck, C. 2010, Diploma thesis, Technische Universität München - Fakultät für Physik / Max-Planck-Institut für Physik Golovin, V. & Saveliev, V. 2004, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 518, 560 Iams, H. & Salzberg, B. 1935, Radio Engineers, Proceedings of the Institute of, 23, 55 Kubetsky, L. A. 1937, Radio Engineers, Proceedings of the Institute of, 25, 421 Burle Industries. 1989, Photomultiplier handbook, Tech. rep., Burle Industries Inc. Hamamatsu Photonics KK Editorial Committee. 2006, Photomultiplier tubes-basics and applications, Tech. rep., Hamamatsu Photonics Otte, A., Dolgoshein, B., Hose, J., et al. 2004, in Nuclear Science Symposium Conference Record, 2004 IEEE, Vol. 2, 1044 1048 Vol. 2 Otte, N. 2007, PhD thesis, Technische Universität München S. Gentile, E. K. & Meddi, F. 2010, IL NUOVO CIMENTO, 8, 999 Wagner, R. 2006, PhD thesis, Technische Universität München Zworykin, V., Morton, G. A., & Malter, L. 1936, Radio Engineers, Proceedings of the Institute of, 24, 351 18