Seminar. Large Area Single Photon Detectors

Transcription

1 University of Ljubljana Faculty of Mathematics and Physics Department of Physics Seminar Large Area Single Photon Detectors Rok Dolenec Mentor: prof. dr. Peter Križan 3. February 2007 Abstract Today a lot of experiments in physics require very sensitive light detectors that must be able to detect even single photons and usually must have a very large active surface. Examples of such experiments are neutrino detectors, where every detected photon counts and very large surfaces must be covered. So far the only light detector capable of reaching those goals was the photomultiplier tube. Recently new approaches, using a solid state photodetector usually called silicon photomultiplier (SiPM) are being explored as possible replacements.

2 Contents 1. Introduction 3 2. Neutrino experiments Neutrinos Detectors Oscillations Importance 5 3. Photomultiplier 5 4. Semiconductor light detectors 6 4. Light Amplifier Approach 7 5. Hybrid Photon Detectors 8 6. Comparison 8 7. Conclusion 9 8. References 9 2

3 1. Introduction Detection of light is an important aspect of many experiments. Light detectors used must fulfil different requirements for different experiments, most demanding of which are the neutrino detectors. Neutrino research could give us better understanding of the Standard Model, dark matter and neutrinos can also be used to observe parts of our universe that are otherwise hidden from us. Neutrino detectors are usually big volumes of water or ice in which neutrinos can interact with atomic nuclei. Interactions are rare, but when they do happen, electrons or muons created emit very faint Čerenkov light. To observe neutrinos we need to detect and measure this light, and for that we need a large number light detectors capable of detecting single photons. Most experiments so far used the photomultiplier tube for this, but many future experiments will only be possible if better and cheaper light detectors can be found. In case of neutrino experiments, the light detectors alone can present as much as 50% of entire experiment s cost. In particle accelerators compact light detectors that are not affected by high magnetic fields are needed, and in all cases the detectors need to be very sensitive to light. Relatively new and quickly developing photodetector usually called silicon photomultiplier is showing many improvements over the photomultiplier tube. It is cheaper, easier to handle and is already showing better light sensitivity than the photomultiplier tube. It is being considered for many experiments and more complex detectors based on it are being developed. In this seminar I will describe neutrino experiments and their demands on light detectors, briefly describe the well known photomultiplier tube and present the possibilities and challenges of new detectors based on SiPM. 2. Neutrino experiments 2.1 Neutrinos Neutrinos are elementary particles that interact with other particles only trough weak interaction. Because cross sections (interaction probabilities) for weak interactions are very small, neutrinos can pass trough a lot of matter before being absorbed or interact with it in any other way. They can pass through the whole Earth without any interaction and are therefore very hard to detect. For some time neutrinos were considered to be massless but it has been proven (by Super-Kamiokande experiment) that neutrinos do have mass, although a very small one (0.04 ev < m [Heaviest ν] < ( ) ev [1]). As all other elementary particles, neutrinos come in three flavours: electron ν e, muon ν µ and tau neutrino ν τ. They can be produced artificially in nuclear reactors and particle accelerators, they are created by nuclear reactions in the stars (Solar neutrinos) and another important source of neutrinos are interactions of cosmic rays with atomic nuclei in Earth s atmosphere (atmospheric neutrinos). Also immense bursts of neutrinos are created in supernovae explosions. We can detect neutrinos by observing the products of their interactions with other particles, usually atomic nuclei in water: Since the cross sections for these interactions are very small, we need to observe very big volumes of water. By detecting typical patterns of Čerenkov light created by electrons and muons, we can identify the events caused by neutrinos and tell which direction they came from and how much energy they had. 2.2 Detectors Surface of the Earth is constantly bombarded by cosmic rays and neutrino detectors are usually located deep underground, so that the rock (or water) above reduces the cosmic ray background. A good example is the Super-Kamiokande (SK) detector, located approximately 1 km underground in the Mozumi mine in Japan. It is a big cylindrical tank filled with tonnes of pure water with its walls covered with photomultiplier tubes that detect Čerenkov light. Photomultipliers used are 50 cm in diameter and were developed by Hamamatsu Photonics specifically for the SK detector. They are capable of detecting single photons and in the final arrangement their photo-cathodes effectively cover 40% of the detector surface. 3

4 Figure 1. A sketch of Super-Kamiokande detector [2] Another important type of neutrino detector consist of a big number of light detectors buried deep under ice or tied to the ocean floor. In this way large volumes of natural detection media can be used. For example, the AMANDA project uses Antarctic ice, which at sufficient depths is exceptionally clear and radiation free. Using a jet of hot water they drill holes in the ice and lower strings with light detection modules on it, so they are positioned at depths from 1 to 1.5 km. When the water in the holes freezes, they get a huge detector mainly used for detecting neutrinos from active galaxies and supernovae and because of that, this type of detectors are also called neutrino telescopes. Even larger neutrino telescopes using volumes up to 1 km 3 are planned or under construction (KM3NET, IceCube). 2.3 Oscillations When observing atmospheric and solar neutrino fluxes, all neutrino detectors observed strange discrepancies. For atmospheric neutrinos it was expected that the flux of muon neutrinos is twice the flux of electron neutrinos: This is due to the most probable decay process of pion, particle that is the direct product of cosmic ray interactions with atmospheric nuclei: or followed by followed by We see that in the end we have two muon neutrinos and one electron neutrino. So every experiment capable of detecting both electron and muon neutrinos should detect twice as much atmospheric muon neutrinos than atmospheric electron neutrinos. But most experiments detected too few muon neutrinos. The difference from expected values is usually reported as the ratio of measured and predicted flux ratios: Because there is less muon neutrinos detected than expected, most experiments measure R to be smaller than 1 (for SK detector R = 0.68 [6]). Similarly for Solar neutrinos they detected only about one third of the expected electron neutrino flux. Both anomalies can now be explained by a process called neutrino oscillations (or mixing). By this process neutrinos of one type can change into neutrinos of a different type. For example, electron neutrinos are created in the Sun and as they travel trough space, they can change to muon or tau neutrinos. By the time they reach Earth, they can be of each of the three flavours with equal probability and so we only detect one third of the expected electron neutrino flux. Similarly atmospheric neutrino measurements are influenced by ν µ ν τ oscillations and 4

5 are also showing zenith angle asymmetry. Neutrinos created in the atmosphere above the detector only have to travel about 15 km while those coming to the detector from below must also travel about km trough Earth and have more time to change flavour. Another evidence for neutrino mixing came from the K2K experiment, where SK was measuring the flux of muon neutrinos created by KEK particle accelerator. From neutrino flux measured at KEK the expected number of neutrinos detected by SK detector 250 km away was estimated to be 151 (if no mixing took place), while SK only detected 108 neutrinos coming from KEK [3]. This result is consistent with those expected from measurements of atmospheric neutrino oscillations. Mixing phenomena are well known from other particles, like kaons, and for mixing to be possible the states involved must have different mass. This suggests that neutrinos too must have some mass, which by now is confirmed by many experiments. 2.4 Importance One of the biggest questions in particle physics is how particles acquire mass and information about the lightest particles, neutrinos, could contribute to answering it. The fact alone that even neutrino has mass is important, because that means that all matter has some mass. Also, if the fourth generation of particles exists, the neutrino would be the first fourth generation particle observed, because the particle with lowest mass would be the easiest to produce in particle accelerators. Neutrinos are also important for cosmology, because neutrino mass could be at least one part of an answer to the dark matter mystery. Neutrino bursts will also alert astronomers of future supernovae and could give us more direct information about supernovae explosions. Because neutrinos interact so little with matter, they come almost unhindered from the innermost regions of the explosion, while most of the visible light comes from decays of radioactive elements created by the supernovae shock wave and even the light from explosion itself is scattered by dense gases. Similarly neutrinos could serve us as a probe to other environments we cannot otherwise see, like the core of our galaxy that is hidden from us by large clouds of dust which visible light or radio waves cannot penetrate. 3. Photomultiplier Figure 2. Photomultiplier tube So far the only light detector capable of reaching the requirements of neutrino experiments was the photomultiplier tube (PMT). PMTs are constructed from a glass vacuum tube which houses a photocathode inside the front surface, several dynodes and an anode. When incident photon hits the photocathode, photoelectron is emitted from it. This electron is then directed towards the first dynode by focusing electrode and accelerated by electric field. When the first dynode is hit by accelerated electron, more electrons are emitted. These are again accelerated towards the next dynode where each of them causes more electrons to be emitted and this process is repeated a couple of times (typically around 10 times). Finally the whole cascade of electrons hits the anode where single photoelectron can be amplified up to 10 8 electrons and the current produced is easily measured. PMTs are able of detecting single photons and can have very large collection area, so with a sufficient number of them, large volumes can be observed. They also feature low noise and good timing resolution. But PMTs also have shortcomings. They require high voltages that are difficult and dangerous to work with and are very fragile (in 2001 while SK detector was under maintenance work, more than half of PMTs imploded and were destroyed because of chain reaction started by a single PMT failure). The production of PMT is fairly complex and still requires a lot of craftsmanship. PMTs are also sensitive to magnetic fields, which causes problems in some applications and can be damaged by accidental illumination. So a cheaper, more robust and easier to produce replacement possibilities are being studied for many future experiments. 5

6 4. Semiconductor light detectors The simplest semiconductor light detector is the photodiode. It is just like a normal diode, only it is constructed so that its depletion region is exposed to incident light. When photons of incident light hit the diode, they can be absorbed and as a result, an electron hole pair is generated. Each absorbed photon thereby generates a small current in the photodiode. But since there is no gain (only one electron hole pair per photon), photodiodes cannot detect small light intensities. If, however, we apply a high reverse bias voltage (100 V or more), electrons and holes can be accelerated enough inside the depletion region to create new electron hole pairs by impact ionization (avalanche effect). Photodiodes specifically designed to operate in this way mode are called avalanche photodiodes (APD) and with them we can achieve gains up to If we increase the bias voltage over the diode breakdown voltage, the diode works in the so called Geiger mode where even larger gains (from 10 6 to even 10 9 ) are possible. Figure 3. The photodiode Figure 4. Avalanche effect The silicon photomultiplier (SiPM, also called multipixel Geiger-mode avalanche photodiode or solid state photomultiplier) is an array of APDs working in Geiger mode joined together on common silicon substrate and working on common load. Each APD (called pixel) is capable of detecting single photons and is typically about 20 µm in size while the typical number of pixels on SiPM is 10 3 /mm 2. In order to minimize the interpixel crosstalk, the pixels must be decoupled from each other, which is achieved with specially designed boundaries between pixels (guard rings) and resistors that are connected to each pixel, which decouple pixels electrically and also limit Geiger discharge. Signal from SiPM is the sum of signals from all pixels, so SiPM can also measure light intensity, which is proportional to the number of pixels fired. The maximum number of photons that can be detected in single pulse (dynamic range of the detector) is limited by the finite total number of pixels on the SiPM, which can be up to 4000/mm 2 [9]. Figure 5. Silicon Photomultiplier [9] Figure 6. Single pixel closeup [9] The SiPM photon detection efficiency (PDE) is: PDE=QE ε Geom ε G Here QE is detector s quantum efficiency the probability that a photon striking the active area gets absorbed and generates electron hole pair (typically QE = depending on the wavelength), ε Geom is the geometrical efficiency that is the fraction of the total SiPM area occupied by active pixel area (typically around 6

7 0.5) and ε G is the probability that a carrier created in active pixel area initiates a Geiger discharge (ε G ~ 0.6). Measured photon detection efficiencies are around 15% to 20% which is similar to photomultiplier tubes, for some wavelengths even better. Fugure 7. A comparison between APD, PMT and SiPM photon detection efficiency [8] This type of detector also has many other advantages: technology to produce it is relatively simple, so detectors would be cheap, it does not require high voltages or cooling to very low temperatures which reduces final application costs and make the operation of them safer and easier, it is insensitive to very high magnetic fields and it can tolerate accidental illumination and it has an excellent timing resolution. Main disadvantage of SiPM is a high dark noise rate that at room temperatures still limits detection of single photons. Dark noise rate originates mainly from the carriers created thermally in sensitive volume and is typically around 1 MHz/mm 2 at room temperature [9]. Dark noise also limits the biggest useful size of SiPM. Typical SiPM area is around 1mm 1mm, the biggest currently available detectors having an area of 3mm 3mm [7]. For bigger detectors the noise rate needs to be reduced by improving pixel production technology. Another disadvantage is untested radiation sturdiness, which is not important for many applications, but for some, like detectors for space telescopes, known behaviour under radiation exposure is crucial. 4. Light Amplifier Approach While SiPM has many excellent features, its main drawback in many applications is that it is very small. The size of 3mm 3mm is obviously too small for neutrino detectors, but SiPM could be very useful with the light amplifier. Figure 8. The SiPM with light amplifier The main idea here is to amplify incoming light before changing it into electronic signal. The design is similar to PMT: incident photons first hit the photocathode, then emitted photoelectrons are accelerated and focused by focusing electrodes. But instead of using a series of dynodes to multiply the signal, we then convert the electrons back to light using a scintillator. Because electrons are accelerated they have more energy and we can get more 7

8 than 1000 photons per initial photon from the scintillator. We can detect this much bigger light flux using SiPM, avalanche photodiode or another PMT. Because here we do not need dynode structure most electronics can be outside the vacuum, which significantly simplifies the production. This design is also less sensitive to magnetic fields than PMTs and if using a fast scintillator, SiPMs excellent timing resolution can be exploited. We could also get a very large angular acceptance. Using specially designed electrodes photoelectrons from spherical photocathode can be focused on a spherical scintillator in the center of the sphere. The idea of using SiPM with light amplifier is very new and only prototypes have been build. But with some development this approach could combine many advantages of PMT with advantages of SiPM. Such detectors could have the same surface as PMTs and even larger PDE while being cheaper, more robust and easier to use. 5. Hybrid Photon Detectors Similar approach is not to convert accelerated photoelectrons back to light, but to detect the photoelectrons themselves using a silicon sensor, which can also be segmented to give spatial information. This design is called Hybrid photon detector (HPD) and has some advantages, but is even more difficult to produce than PTM because silicon sensor must be inside the vacuum. The HPD can also have large angular acceptance by putting the silicon sensor in the middle of sphere coated with photocathode. This way big detectors (40 cm in diameter) with angular acceptance of ±120 could be build [10]. 6. Comparison Figure 9. Simple Hybrid Photon Detector Figure 10. Typical light detection efficiencies of PMT, SiPM and HPD The SiPM is still in development, but already it is a very attractive replacement for the PMT. Even with the light amplifier it is easier to produce and thus cheaper and its PDE is already very near or better than the PMT quantum efficiency. With some further development the SiPM PDE will be improved even further and it can be optimized for different wavelengths. Other technologies like the HPD could be better for some special applications, but for most its high cost makes it less attractive than the PMT. 8

9 PMT APD SiPM HPD Gain QE (PDE) > 20% 80% < 20% > 20% Treshold sensitivity 1 ph.e. 10 ph.e. 1 ph.e. 1 ph.e. Timing 100 ps a few ns 30 ps 100 ps Voltage 1 kv 100 V < 50 V 20 kv Complexity high medium low very high Table 1. Comparison of most important light detectors [9] 7. Conclusion Most widely used light detector, the photomultiplier tube, is now almost a 100 years old design. It is still being constantly improved but new type of detectors are becoming available. Detectors based on SiPM outperform PMTs in many ways and also promise to be a lot cheaper. This is especially important for neutrino experiments, where light detectors alone can present as much as 50% of entire experiment s cost. Properties of SiPM alone, like very high magnetic field tolerances, light weight and low power consumption also make them very attractive for applications like Čerenkov counters in particle colliders, space telescopes and medical applications. 8. References [1] W.-M. Yao et al., Journal of Physics G 33, 1 (2006) ( [2] The Super-Kamiokande Collaboration: The Super-Kamiokande Detector, Nucl. Instrum. Meth. A501(2003) ; [3] K2K Homepage: [4] IceCube Neutrino Observatory Homepage: [5] Peter Križan, Detection of neutral particles [6] Anže Zupanc: Nevtrinske oscilacije [7] B.Dolgoshein, Large Area Silicon Photomultipliers: Performance and Applications, 4 th ICNDP, Beaune ( [8] B.Dolgoshein, Silicon Photomultipliers in ParticlePhysics: Possibilities and Limitations ( [9] P. Buzhan, B. Dolgoshein: An Advanced Study of Silicon Photomultiplier, ICFA Instrumentation Bulletin [10] Christian Joram, Advances and trends in large-area Hybrid Photon Detectors (HPD) [11] D. Renker, Silicon Photomultipliers, Beaune 2005 [12] E. Lorenz, MPI-Munich,Germany, Exotic Ideas in Photon Detectors 9