Hybrid Avalanche Photo Detector for Belle II ARICH detector

University of Ljubljana Faculty of Mathematics and Physics Physics department Seminar I Hybrid Avalanche Photo Detector for Belle II ARICH detector Author: Mitja Predikaka Suprevisors: dr. Rok Pestotnik December 27, 214 Abstract In this paper, a novel photon detector for proximity-focusing Aerogel Ring Imaging Cherenkov detector in Belle II spectrometer is presented. It will be used for separation of kaons from pions over the most of their momentum spectrum and to discriminate between pions, muons and electrons below 1 GeV/c. Contents 1 Introduction 1 2 Introduction to Particle Identification (PID) 1 2.1 Cherenkov radiation......................................... 2 2.2 PID.................................................. 2 3 Proximity-focusing Aerogel Ring Imaging Cherenkov detector 3 3.1 Aerogel Radiator........................................... 4 3.2 Photon Detector........................................... 5 3.2.1 From photodiode to hybrid avalanche photodiode..................... 5 3.2.2 Hamamatsu Hybrid Avalanche PhotoDetector (HAPD)................. 6 4 Properties of Hamamatsu HAPD 6 4.1 Position resolution.......................................... 6 4.2 Gain.................................................. 7 4.3 Pulse height spectrum........................................ 7 4.4 Background contribution....................................... 8 4.5 Quantum Efficiency (QE)...................................... 8 4.6 Magnetic field immunity....................................... 9 5 Particle Identification (PID) by proximity-focusing RICH with aerogel radiator 1 6 Conclusion 1

1 Introduction Standard model (SM) of particle physics has proved itself as very successful theory. At the time of formulation it has explained many of experimental results and has predicted new particles and processes that were later found. Its biggest success is combining two of the four known fundamental interactions into unified electroweak theory, and the development of quantum chromodynamics, which describes the strong interaction. Nevertheless there are facts and phenomena that cannot be explained via SM, like gravitational theory, matter antimater asymmetry, neutrino mass, and many more. These arguments do not mean that SM is wrong theory, only that it is part of a something bigger. All of the above arguments suggest that there are processes outside of SM that we do not yet know. Those processes, mechanisms and possible new particles are called New physics, and are currently the main focus of experiments in physics of high energies. One of the ways to find New physics is to search extremely rare events at energies around few GeV with high precision. Belle II collaboration has decided to search so called New physics in measurements of rare B, D and decays. To do this efficiently, they are currently upgrading KEK B collider to SuperKEKB collider, that will have very high luminosity to provide needed statistics. Furthermore they are building new Belle II spectrometer to log all interesting events, for which they need extremely accurate: vertex and tracking system, particle identification system, and energy measuring system. Hereinafter, I will focus on a specifically for Belle II spectrometer on SuperKEKB collider developed Aerogel Ring Imaging Cherenkov detector for particle identification and its novel type of photon detector, that has been developed for use in high energy physics. Figure 1: Schematic view of SuperKEKB collider (left) and Belle II spectrometer (right) 2 Introduction to Particle Identification (PID) Particle identification (PID) is a process of identifying particle type based on the information, that is left in a detector upon particle traversing. It is one of the fundamental requirements of all experiments in high energy physics. Therefore, this section will cover the basics of PID via Cherenkov radiation. Reader can learn more about PID in general in [1]. 1

2.1 Cherenkov radiation When a charged particle traverses the medium faster than the speed of light in that medium Cherenkov radiation (polarized electromagnetic radiation) at a characteristic Cherenkov angle is emitted. Relation between that angle and a speed of particle is given as cos C = 1 n, (1) where C is Cherenkov angle, is speed of particle divided by speed of light and n is the refractive index of a surrounding medium. Molecules in close proximity of particle are momentarily polarized, and after its passage those molecules return to their initial state by emitting photons, which can interfere destructively or constructively, depending on the particle speed. Constructive interference of photons caused by particles traversing the matter faster than speed of light, is called Cherenkov radiation. Graphical representation is shown in Fig. 2. Figure 2: Polarization of matter by charged particle traversing the matter. Left: particles slower than electromagnetic waves induce symmetrically arranged dipoles around their path no net dipole moment and no Cherenkov radiation. Center and right: particles faster than electromagnetic waves break symmetry causing non-vanishing dipole moment radiation of Cherenkov photons at a characteristic angle C. Reader can learn more about Cherenkov radiation from [2] and [3] (both in Slovenian). 2.2 PID Now we can check dependance of Cherenkov angle on particle momentum for different particles. Fig 3 shows that Cherenkov angles for kaons and pions are well separated in a range of energies up to 4 GeV/c in the same medium. But because of the measurement errors we do not get delta function at specific Cherenkov angle, but rather a distribution of Cherenkov angles (Fig. 4) at a specific particle momentum. Measurement errors arise from uncertainty of emission point, finite position resolution of a photon detector, variation of the refractive index of medium and its imperfections and inhomogeneity, and from error in track parameters determined by the tracking system that are extrapolated to the radiator. From here on, I will only focus on the first two. Emission point uncertainty ( emp of a photon detector ( pix )are )andfinitepositionresolution emp = d l p 12 sin C cos C, (2) pix = a l p 12 cos2 C, (3) where d, l and a are thickness of a medium, distance from medium to photon detector and size of a photon detector, respectively. Now if we sum above errors in quadrature we get single photon resolution ( ), which is directly connected to Cherenkov angle resolution ( ) that will be used in PID, = p Ndet, (4) 2

6.3 5 C (K) C ( ).25 4 qchradl Entries 3.2 2.15 K Hn=1.46L p Hn=1.46L 1 K Hn=1.56L K p Hn=1.56L p 1 2 3 4.24.26.27.28.3.32.34 momentum HGeVêcL qchradl Figure 3: Cherenkov angle as a function of particle momentum for kaons and pions in medium with n =1.46 and n =1.56. Figure 4: Cherenkov angle distribution for kaons and pions at approximately 3.5 GeV/c. C (K) C ( ) = 3 mrad in a medium with n 1.5. where N det is number of detected photons by photon detector. From physical requirements of the experiment, we want to separate kaons from pions with at least 4, therefore 4 < C (K) C ( ), (5) where C (K) C ( ) is the difference in measured Cherenkov angles for different particles (Fig. 4). 3 Proximity-focusing Aerogel Ring Imaging Cherenkov detector This section will focus on PID system in the forward in the direction of electron beam endcap region of spectrometer. PID system specifically designed for Belle II has to comply with physical and detector requirements. Physical requirements are: - separation between kaons and pions up to 4 GeV/c with at least 4, - and discrimination between pions, muons and electrons below 1 GeV/c. And the detector requirements are: - immunity to 1.5 T high magnetic field, - high tolerance for neutron and gamma radiation, - and system has to fit in a 28 cm wide space. Type of system that fulfills all of the above criteria is called proximity-focusing Ring Imaging Cherenkov (RICH) detector with aerogel as a radiator (ARICH). Schematics are shown in Fig. 5. Proximity-focusing ARICH, consists of following elements: - an aerogel radiator where charged particles emit Cherenkov photons, - an expansion volume where Cherenkov rings gain size, - an array of position sensitive photon detectors where Cherenkov rings are registered, - and readout electronics. 3

aerogel photon detector Cherenkov photons charged particle n 1 n 2 n 1 <n 2 d l Figure 5: Schematics of proximity-focusing ARICH detector [4]. Figure 6: Schematics of proximity-focusing ARICH with multiple radiators that improve separation capability [4]. 3.1 Aerogel Radiator Aerogel radiator is the medium, in which charged particles emit Cherenkov photons. Number of Cherenkov photons emitted per unit length can be written as: d 2 N d dx = 2 z2 1 1 2 2 n 2 = 2 z2 sin 2 ( ) 2 C, (6) where z and are particle charge and fine-structure constant, respectively. For single charged particle, integrated over the peak of QE photocatode spectrum (3 nm 5 nm), this gives dn dx 611 photons sin2 C cm, (7) from where it can be seen that charged particles produce some characteristic number of photons per unit length. Cherenkov angle for pions at 3.5 GeV/c is approximately 37 mrad (with n 1.5), so this gives maximum of 55 emitted photons per centimeter of radiator. If we could detect all of the produced photons, measurements would be really accurate. But we have to take into an account that we detect less photons Z d N det e a t dx Z 5nm 3nm q ( )T ( ) d2 N d dx d, (8) where N det is number of detected photons, e is the single photoelectron detection efficiency, a is the active surface fraction of the photon detector, t =.9 is the fraction of the surface covered with photon detector modules in a particular tiling scheme, d is radiator thickness, q ( ) is quantum efficiency of photon detector, and T ( ) is aerogel transmission. Specifications regarding the selected photon detector for proximityfocusing RICH detector are gathered in Table 1. Taking into account all factors of equation 8, we will detect only around 1% of all emitted Cherenkov photons. To maximize detection of photons, aerogel should have high transsmitance T ( ), inordernottolose photons inside it via Rayleigh scattering or absorption. From eq. 7 and 2 we see, that both are connected with the thickness of a radiator. Therefore, we have to find optimal thickness, that will yield appropriate number of photons, without loosing Cherenkov angle resolution due to emission point uncertainty. Emission emp point uncertainty as a function of aerogel thickness d contributes 4.9 mrad/cm d, andtheoptimal thickness of a radiator would be 2 mm [4], but that does not yield enough photons for sufficient Cherenkov angle resolution ( ). The question was how to get enough photons per track without loosing Cherenkov angle resolution. This was solved by stacking together two layers of aerogel radiator with appropriate refractive indices in focusing configuration (Fig. 6). As seen in Fig. 7 focusing configuration improves single photon resolution from = 2.7 mrad to = 14.3 mrad, with the number of detected photons remaining the same. 4

N 8 6 195. / 116 P1 7289. P2.374 P3.1428E-1 P4 74.49 P5 884.4 N 6 4 2467. / 116 P1 5495. P2.2965 P3.272E-1 P4 85.32 P5 796. 4 2 nf= 1.27 nb= 1.5 2 nf= 1.64 nb= 1.36.1.2.3.4.5 θ c (rad).1.2.3.4.5 θ c (rad) Figure 7: Number of Cherenkov photons as a function the Cherenkov angle from beam test with MCP PMT as photon detector. Left: focusing configuration of two 2 cm thick aerogel radiators with refractive indices of 1.46 and 1.56. Right: 4 cm thick homogeneous aerogel radiator [4]. 3.2 Photon Detector Initially, there were three proposed photon detectors for proximity-focusing RICH detector: Silicon Photo- Multipliers (SiPM) with light guides, MicroChannel Plate PhotoMultiplier Tubes (MCP PMT) and Hybrid Avalanche PhotoDetector (HAPD). Out of them, SiPM proved to be susceptible to neutron radiation damage, and photocatode of MCP PMT did not cope well with aging (long exposures to photons). Therefore, HAPD (right side of Fig. 1) produced by Hamamatsu Photonics was chosen for proximity-focusing RICH photon detector. 3.2.1 From photodiode to hybrid avalanche photodiode Photodiode is the simplest semiconductor photon detector based on silicon p n junction (Fig. 8). Diffusion of charge carriers on junction creates a depleted region with weak electric field. If particle like photon generates electron hole pair in that region (or near the edge) electric field pushes electrons to the n doped layer, and holes to the p doped layer. That movement of charge (current) is too weak to be measured. So the lack of internal amplification (gain) makes photodiodes applicable only when one expects more than 1 photons, each producing an electron-hole pair.. Conduction band Valence band Band gap Photon p-type Depleted region n-type Hole Electron Figure 8: Detection of light in silicon photodiode [5]. Figure 9: Hybrid avalanche photodiode [6]. 5

For detection of weaker photon intensities, as it is our case, one can add reverse bias on photodiode, which increases internal electric field enough, so that electrons can receive enough kinetic energy for further ionization. Photodiode with reverse bias is called an avalanche photodiode (APD). Internal amplification (avalanche gain) is limited by breakdown voltage voltage at which multiplication by holes becomes considerable, and the multiplication is no longer controlled up to around factor 1. This is still not enough for single photon detection. To achieve higher gain one can combine vacuum technology with semiconducting photodetectors. By replacing dynode and anode structure of classical photomultiplier tubes with high electric field and avalanche photodiode (Fig. 9), respectively, we get HybridAPD. When a photon hits photocatode it generates photoelectron via photoeffect. Electron is then accelerated by high electric field, so when it hits avalanche photodiode at the bottom of HybridAPD, it has enough energy to generates several 1 primary pairs via ionization. Average number of generated pairs defines the so called bombardment gain. Next step in multiplication is the same as with APD. So the total gain of HybridAPD is convolution of bombardment gain and avalanche gain, which is enough for single photon detection capabilities. 3.2.2 Hamamatsu Hybrid Avalanche PhotoDetector (HAPD) Hamamatsu HAPD that will be part of proximity-focusing RICH detector with aerogel radiator, is a squareshaped HybridAPD (Fig. 1), that consists of super bialkali photocatode, vacuum region with high electric field (typical bias is around 7 kv ), and four APD chips, that are further pixelated in 6 6 pixels. Flawless production of big APD chips is very difficult and expensive. Therefore, it is easier and cheaper to make HAPD from four smaller APD chips, than from one 12 12 pixels big APD. Major specification are gathered in Table 1. Super bialkali photocathode photon ~7 kv ~2 mm e- Pixelated APD Figure 1: Schematic view (left) and picture (right) of HAPD. Adapted from [4]. package size (mm 2 ) 73 73 weight (g) 22 sensitive area (%) 64 peak quantum efficiency (%) 3 number of pixels 144 (36 4chips) bombardment gain 15 pixel size (mm 2 ) 4.9 4.9 avalanche gain 3 capacitance (pf) 8 total gain 45 1 3 single photon detection efficiency e.8 active surface fraction a.67 Table 1: HAPD specifications from [4] and [7]. 4 Properties of Hamamatsu HAPD 4.1 Position resolution Finite pixel size gives, in addition to emission point uncertainty, another contribution to single photon resolution. For a pixel size of a =4.9mm, we get finite position resolution (eq. 3) of HAPD =6.6mrad, 6 pix

when distance from the radiator to photon detector is l = 19 cm. 4.2 Gain Mean energy required to produce an electron-hole pair in silicon is approximately W = 3.66 ev [8]. In theory, primary photoelectron with energy E could ionize E/W secondary electron-hole pairs. From this, we can see linear connection between bombardment gain and photocatode voltage, as seen in Fig. 11. Exponential growth of an avalanche of secondary electron-hole pairs is a bit harder to see, but electrons that gain at least 3.66 ev of energy on their mean free path can generate another pair and so on. Therefore, avalanche gain has exponential dependance on reverse voltage (Fig. 12). Electron Bombardment Gain 2 15 1 Chip A Chip B Chip C Chip D Avalanche Gain 1 8 6 4 Chip A Chip B Chip C Chip D 5 2 2 3 4 5 6 7 8 Photocathode Applied Voltage @-kvd Figure 11: Linear dependance of bombardment gain on photocatode voltage for typical HAPD. 1 2 3 4 Reverse Bias Voltage @VD Figure 12: Exponential dependance of avalanche gain on reverse bias voltage for typical HAPD. 4.3 Pulse height spectrum Response of HAPD to short light pulses with wavelength of 47 nm can be seen from Fig. 13. Excellent photon counting ability is evident as peaks corresponding to different number of detected photons are finely separated. The reason behind the background between peaks is backscattering of photoelectrons (Fig. 14), which leave only partial ionization in silicon. 1 Frequency 8 6 4 1pe 2pe 3pe 4pe Photocathode Voltage: -8 kv Reverse Bias Voltage: 348 V Guard Voltage: 175 V Light Source: LED 47 nm, 2 khz... 2 1 2 3 4 Output Pulse Height HADC chl Figure 13: Pulse height spectrum where peaks corresponds to different number of detected photons. 7

4.4 Background contribution From the position of incident Cherenkov light on photon detector we can determine Cherenkov angle and fill histogram it. On the right side of Fig. 14, we see that we do not get only one gaussian peak per incident charged particle of selected momentum (as in Fig. 4), but also additional background. Background can be explained by Cherenkov light that is reflected from APD surface and then produces photoelectron, by photoelectron backscattering, Cherenkov light from window, and internally reflected Cherenkov light. All this contributions are schematically shown in Fig. 4. photon window photoelectron photocatode APD Cherenkov light reflected from APD surface 9 8 7 AEROGEL Cherenkov light Photoelectron backscattering Cherenkov light from window Number of events 6 5 4 3 Internal reflections 2 1 2 mm 2 mm.1.2.3.4.5.6.7.8.9 1 Cherenkov angle [rad] Figure 14: Sources of additional background peaks. For additional info look into text. 4.5 Quantum Efficiency (QE) Quantum efficiency (QE) is the number of photoelectrons emitted from the photocathode divided by the number of incident photons. Therefore, photocatodes with higher QE will emit more photons and as a result, achieve better Cherenkov angle resolution (eq. 4). QE peak of 3% at 36 nm (as seen in Fig. 15) has been achieved for typical HAPD with Super Bialkali photocatode it is important to note, that QE peak coincides with wavelengths of Cherenkov radiation. In comparison, photocatodes of classical photomultipliers have highest QE with Ultra Bialkali photocatodes up to 43%. 3 5 TPMHB775EA Quantum Efficiency @%D 25 2 15 1 5 2 3 4 5 6 7 Wavelength @nmd QUANTUM EFFICIENCY (%) 45 4 35 3 25 2 15 1 5 2 Ultra Bialkali Super Bialkali Current Bialkali 25 3 35 4 45 5 55 6 65 7 WAVELENGTH (nm) Figure 15: QE measured for typical HAPD. Figure 16: QE of photocatodes from classical photomultiplier tubes for Bialkali, Super Bialkali and Ultra Bialkali technology [9]. 8

4.6 Magnetic field immunity Unlike classical photomultiplier tubes that cannot operate in high magnetic fields due to bending of electron trajectories, HAPD performance is improved by it. Magnetic field that is perpendicular to the entrance window of HAPD causes electrons to circulate along the magnetic field lines (with radius around 3 nm [1]) and consequently they fall where they should. Therefore, magnetic field decreases distortion due to a nonuniform electric field at the edge (Fig. 17) and eliminates photoelectron backscattering cross-talk (Fig. 2). Figure 17: One dimensional scan of HAPD with B =T (left) and B =1.5 T (right) [4]. Projection X of row 6 1 2 3 4 5 6 7 8 9 1 11 12 Figure 18: One dimensional scan of row 6 done on a testing HAPD, measured on IJS. Numbers from 1 to 12 denote HAPD pixels. We can see the lack of signal on pixel position 7, which can happen because of a damaged pixel or readout electronic. On one dimensional scan without magnetic field (left half of Fig. 17 and Fig. 18) two red coloured peaks appear near the left edge. This means that when light incident position is at the edge incident electrons ionize first two pixels of APD. Reason behind this is a nonuniform electric field on the edges. Fig. 19 show trajectories of electrons in HAPD with B =T. 9

Figure 19: Electron trajectories in B =T [1]. 1 1 9 1 3 9 1 3 8 8 1 2 1 2 7 7 6 6 1 1 5 5 4 2 3 4 5 6 7 8 1 4 2 3 4 5 6 7 8 1 Figure 2: Electron backscattering with B =T (left) and B =1.5 T (right) [4]. 5 Particle Identification (PID) by proximity-focusing RICH with aerogel radiator As stated in section 3 we want to separate kaons from pions with at least 4. Contributions to single photon emp resolution from emission point uncertainty (eq. 2) and finite position resolution (eq. 3) give = 19.6 mrad (for d =4cm) and pix =6.6 mrad, both at l = 19 cm. This gives = 2.7 mrad. Taking into account 2 detected photons we get Cherenkov angle resolution (eq. 4) of = 4.6 mrad. Difference in Cherenkov angle between kaons and pions from Fig. 4, C (K) C, ( ) = 3 mrad at 3.5 GeV/c gives us 6.5 separation. Therefore, precision of measurement is higher than that from physical requirements. On this point it should be stated, that actual separation is a bit lower, because we did not take into account errors that arrive from other sources as stated in section 2. 6 Conclusion In the search of very rare events at energies around few GeV particle identification plays an important role. Therefore, the conditions on the novel HAPD photon detector of the proximity-focusing aerogel RICH detector were tough. HAPD has to provide accurate measurements of Cherenkov angles to separate kaons and pions over the most of their momentum spectrum with only around 2 detected photons, work in a high magnetic field and be resistant to high neutron and gamma fluxes. Even though it was not the best in all specifications, 1

its overall performance made it best choice in given working environment. To conclude, proximity-focusing aerogel RICH detector will be employed for particle identification system in the forward end-cap of the Belle II spectrometer, with HAPD for its photon detector. The beamtest results and the detector simulations have shown that the designed detector will have an excellent kaon identification efficiency of more than 95% over wide range of momentum at a rather low pion misidentification probability of 1%. References [1] C. Lippmann, Particle identification, Nuclear Instruments and Methods in Physics Research A, vol. 666, pp. 148 172, Feb. 212. [2] D. Grahelj, Sevanje Čerenkova, 21. http://mafija.fmf.uni-lj.si/seminar/files/29_21/ sevanje_cerenkova.pdf [Online; accessed 27. December 214]. [3] M. Bajić, Detektorji sevanja Čerenkova, 211. http://mafija.fmf.uni-lj.si/seminar/files/ 211_212/detektorji_cerenkova.pdf [Online; accessed 27. December 214]. [4] T. Abe, I. Adachi, K. Adamczyk, S. Ahn, H. Aihara, K. Akai, M. Aloi, L. Andricek, K. Aoki, Y. Arai, and et al., Belle ii technical design report, arxiv:111.352v1 [physics.ins-det], Nov. 21. [5] P. Križan and S. Korpar, Photodetectors in Particle Physics Experiments, Annual Review of Nuclear and Particle Science, vol.63,no.1,pp.329 349,213. [6] S. Korpar, Status and perspectives of solid state photon detectors, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, vol. 639, no. 1, pp. 88 93, 211. [7] R. Pestotnik, Aerogel RICH for the Belle II spectrometer, Lecture on 13th Vienna Conference on Instrumentation, February 213. [8] F. Scholze, H. Rabus, and G. Ulm, Mean energy required to produce an electron-hole pair in silicon for photons of energies between 5 and 15 ev, Journal of Applied Physics, vol.84,no.5,pp.2926 2939, 1998. [9] K. K. H. Photonics, Photomultiplier Tube Series UBA and SBA, 28. http://shpat.com/docs/ hamamatsu/uba_sba_tpmh135e2.pdf [Online; accessed 27-December-214]. [1] L. Šantelj, Trajectories of electrons in HAPD, October 28. Private communication. 11