The Fluorescence Detector of the Pierre Auger Observatory

Nuclear Physics B (Proc. Suppl.) 165 (2007) 37 44 www.elsevierphysics.com The Fluorescence Detector of the Pierre Auger Observatory V. Verzi a for the Pierre Auger Collaboration b a Sezione INFN di Roma Tor Vergata, via della Ricerca Scientifica 1, 00133 Roma, Italy b Observatorio Pierre Auger, av. San Martine Norte 304, 5613 Malargue, Argentina Three of the four fluorescence detectors of the Pierre Auger Observatory operate stably. The main detector features, the absolute calibration and the atmospheric monitoring will be described. The detector performance will be also described, with particular attention to the determination of the shower longitudinal profile and primary energy. 1. INTRODUCTION The Pierre Auger Observatory has been designed to study the ultra-high energy cosmic rays (UHECR) in the GZK cutoff region [1,2] with an unprecedent statistics and with low systematic uncertainties. The observatory will consist of two sites, one in each hemisphere. The southern detector [3] is currently under construction close to the city of Malargue in Argentina, in a site called Pampa Amarilla. The detector is made by an array of water-cherenkov tanks [4] spread over a surface of approximatively 3000 Km 2 and by four fluorescence detectors looking to the array. At the moment three fluorescence detectors are in operation. Two of them, Los Leones and Coihueco, have been collecting data since January 2004, while Los Morados started in March 2005. The fourth site at Loma Amarilla is under construction and will be in operation at the beginning of 2007. The main feature of the fluorescence detector (FD) is the capability to measure the longitudinal energy deposit allowing a nearly calorimetric measurement of the cosmic ray energy. In contrast, the surface detector (SD) can only measure the shower size at ground and the extrapolation of this measurement to the primary energy is based on Monte Carlo simulations which, at the GZK energies, are affected by large systematic uncertainties. Unfortunately the FD has a duty cycle of only about 10%, as it can only operate during clear nights with little moonlight, while the SD system has a potential duty cycle of 100%, very important for the detection of the low flux of UHECR. The 10% of cosmic ray showers, detected simultaneously by both detecors, are called hybrid events and are of a fundamental importance for the achievement of physics objectives of the Pierre Auger Observatory. In fact the FD calorimetric measurements of those hybrid atmospheric showers will be used to calibrate the SD energy estimator, allowing a substantial reduction of the systematic uncertainties. In the following sections the Auger fluorescence detector and the associated instruments for calibration and atmospheric monitoring will be rewieved. The various steps of the reconstruction of the fluorescence event data will be also described with particular emphasis to the determination of the longitudinal shower profile and primary energy. 2. THE FLUORESCENCE DETECTORS In each of the four sites the fluorescence detectors consist of six telescopes, each located in independent bays, overlooking separate volumes 0920-5632/$ see front matter 2006 Published by Elsevier B.V. doi:10.1016/j.nuclphysbps.2006.11.007

38 V. Verzi / Nuclear Physics B (Proc. Suppl.) 165 (2007) 37 44 of air. Figure 1 shows a schematic cross-sectional view of one fluorescence telescope. The telescope is built with a Schmidt optics allowing the elimination of the coma aberration: a circular diaphragm of radius 1.10 m is located at the center of curvature of the spherical mirror. An UV transmitting filter placed in the entrance aperture reduces the background light and provides protection from outside dust. A series of Schmidt corrector elements are located just inside the UV filter. The light is focused by a large 3.5 m 3.5 m spherical mirror with curvature radius of 3.4 m onto a spherical camera, which hosts an array of 440 photomultipliers (PMT) with photocatode of exagonal shape. The camera field of view is 30 elevation 28.6 azimuth. The angular size of the spot from spherical aberration is 0.5 which is 1 3 of the pixel size and it is practically independent from the incident direction. each photomultiplier has been surrounded by an hexagonal set of flat reflecting surfaces. The basic element of the light collector is a Mercedes star fixed at the vertex of three adjacent pixels. The signal coming from each telescope is sampled at 10 MHz with 12 bit resolution. The first level trigger system performs a boxcar running sum of ten samples. When the sum exceeds a threshold, regulated to keep the trigger rate close to 100 Hz, the trigger is fired. The second level trigger scans the camera every microsecond for patterns of fired pixels, that are consistent with a track induced by the fluorescence light from shower. A more sophisticated software trigger is also implemented. It selects shower candidates and performs a fast reconstruction of shower geometry. For candidates considered good showers, a signal is sent to the central data acquisition in order to perform the hybrid trigger [5]. The observed rate of good showers, satisfying this trigger, is about 5 h 1 per telescope. 3. DETECTOR CALIBRATION A necessary condition for a precise measurement of the cosmic ray energy is a precise absolute calibration of the detector. The absolute telescope calibration consists in the determination of the conversion factor between the PMT ADC counts and the light flux at the telescope aperture. This conversion factor includes all the detector components (optical filter transmittance, mirror reflectivity, PMT quantum efficiency and gain, etc...). The absolute calibration is performed few times per year. The stability of the telescopes is monitored every night during detector operation. Figure 1. Sketch of a fluorescence detector telescope. From right to left one can see the spherical mirror, the camera and the aperture system. To maximize light collection and guarentee a smooth transition between adjacent pixels, 3.1. Absolute calibration Ideally, the end-to-end FD absolute calibration should be performed using an absolutely calibrated light source providing light of the same wavelength and distribution as it is measured in cosmic ray showers. For the Auger fluorescence detectors this means that the light source should inject uniformly distribuited photons on the telescope aperture for any given direction with wavelengths between 300 and 400 nm.

V. Verzi / Nuclear Physics B (Proc. Suppl.) 165 (2007) 37 44 39 In order to match those requirements, the technique adopted by the Auger collaboration is based on a portable 2.5 m diameter drum-shaped light source, which mounts on the exterior part of the FD apertures, providing a pulsed photon flux of known intensity and uniformity across the aperture and simultaneously triggering all the PMTs in the camera. The light source is mounted in a reflector cup placed in the center of the drum front surface and illuminating the interior of the cylindrical drum. In order to isotropise the light, the internal surface of the drum is covered with Tyvek, a material diffusively reflective in the UV, while the front face of the drum is a thick Teflon sheet, which transmits light diffusively. Photons/(ADC count) 7 6 5 4 3 Before flat-fielding After flat-fielding After 3 months 0 100 200 300 400 Pixel Number Figure 2. Photon per ADC count measured for all 440 pixels of bay 5 at Los Leones. The blue line shows the conversion factors as commissioned, the black line immediately after calibration and the red line shows the typical variation observed at next calibration. At the present time, the FD is calibrated at only one wavelength using UV leds emitting in a narrow band around 375 nm. The light source is absolutely calibrated using UV enhanced Si photodetectors calibrated at NIST. Modifications to the present system are under development to allow multi-wavelength calibration using a Xenon light source. The uniformity of light emission from the drum surface has been verified at the level of 2% using a CCD viewing the drum from a distance of 14 m. The uniformity of the light emission for different photon emission angles has been also verified changing the angle between the drum axis and the CCD axis. Figure 2 shows the effects of the drum calibration on the 440 pixels of one FD camera. The typical sensitivity for FD pixels is about 5 photons per ADC count. The systematic uncertainty on the calibration is estimated to be 12% [6]. 3.2. Relative calibration Changes in the properties of any component of the FD telescope between successive absolute calibration is tracked in time by routine relative calibration measurements. These measurements determine the response of each pixel of the telescope to light pulses from either a LED source or a xenon flash lamp. The light is brougth via optical fibers to three different diffuser groups for each telescope: at the center of the mirror to illuminate the camera; along the lateral edges of the camera body facing the mirror; facing two reflecting Tyvek foils mounted on the inner side of the telescope doors. The relative calibration constants are measured with respect to a given relative calibration run taken within one hour after the absolute calibration measurement. The short and long term stability of the telescopes have been measured [7] but, at the present time, they have not been yet included in the reconstruction chain. This corresponds to a systematic uncertainty of 5% which is already included in the 12% uncertainty on the FD absolute calibration constants. 4. ATMOSPHERE MONITORING The fluorescence detection technique uses the earth atmosphere as a calorimeter medium. A precise knowledge of the interaction processes of the fluorescence light in the atmosphere is required. The Rayleigh scattering occurs with the atmospheric molecules. It is a very well known process and its determination requires only the measurement of the atmospheric density as a function of the height. In contrast the aerosol scattering

40 V. Verzi / Nuclear Physics B (Proc. Suppl.) 165 (2007) 37 44 process is unknown a priori and should be monitored. The aerosol scattering cross section and the aerosol content of the atmosphere, in the form of clouds, dust, smoke and other pollutants, needs to be well characterized. Finally, a precise knowledge of the atmospheric conditions is also needed to predict the correct value of the fluorescence yield along the shower axis, being it dependent on temperature and pressure. 4.1. Atmospheric profiles monitoring The atmospheric conditions at the observatory are monitored using metereological radio soundings and ground based weather stations [8]. The radiosondes are launched above the site on helium-filled balloons providing data up to 25 km above the sea level. In addition, information from public databases of atmospheric radio soundings launched from sites about 500 km far from Malargue have been used to develop a monthly model of the atmospheric profiles above the Auger site. Those models are currently used in the Auger reconstruction and simulation of cosmic air showers. 4.2. Aerosol monitoring The aerosol content of the atmosphere can be variable on a short time-scale and can be not uniform in the area enclosed by the observatory. For that reasons the aerosol is characterized into 5 different regions and, within each region, in vertical slices of 200 m thickness, up to height of 10 km. One of the main instrument dedicated to the aerosol monitoring is the elastic backscatter LI- DAR [9], which will be placed near each of the FD sites. Currently two of these LIDARs are operational. The LIDAR consists of a steerable pulsed UV laser. The light is detected by photomultipliers placed at the foci of parabolic mirrors. During FD operation the LIDAR system scans the sky providing aerosol measurements every hour. Moreover for a detected shower, the FD sends a signal to the LIDAR providing the estimated shower-detector plane. Received this signal, the LIDAR starts a scan of the atmosphere along the shower-detector plane. As a part of the LIDAR program a Raman LIDAR detector has been installed in one LIDAR site providing a more accurate reconstruction of the aerosol transmission. The other main aerosol monitoring instrument is the Central Laser Facility (CLF) [9,10]. It is placed in the middle of the Auger observatory at distances ranging from 26 to 39 km from the FDs. It consists of a 355 nm steerable laser providing pulses with a width of 7 ns and up to energy of 7 mj. The scattering of photons out the laser firing direction produce a signal in the FDs very similar to the shower signal. Hundreds of laser shots are fired every hour during FD operation. An analysis of the FD vertical laser tracks allows to determine the vertical aerosol optical depth as a function of the altitude (VAOD(h)). Figure 3 shows the distribution of the measured VAOD between FD altitude and 4500 m above the sea level. Number 70 Los Leones Coihueco 60 50 40 30 20 10 0 0 0.05 0.1 0.15 0.2 VAOD at 4500m a.s.l. Figure 3. Distribution of VAOD between FD altitude and 4500 m above the sea level. Vertical laser shots fired by the CLF and detected in two FD sites have been used for the VAOD determination. The two distributions are different because of the different altitude of the two FD sites.

V. Verzi / Nuclear Physics B (Proc. Suppl.) 165 (2007) 37 44 41 The wavelength dependence of the aerosol scattering cross section is monitored using the Horizontal Attenuation Monitor (HAM) [9]. The HAM consists of a light source and a receiver located in two different FD sites and measures the attenuation length at near ground level at several wavelengths. The Aerosol Phase Function Monitors (APFs) [9] are designed to measure the aerosol differential scattering cross section. The measurement is done firing an horizontal beam of light across the front of an FD and than producing a track with a wide range of scattering angles from the beam direction. Finally star monitors [9] are used to measure the total optical depth from ground to the top of the atmosphere and a sky map of the cloud distribution is obtained by finely pixelated infrared cameras [9] which are sensitive to the temperature difference between the clouds and the clear sky. 5. FD PERFORMANCE The first step of the shower reconstruction is the determination of the shower axis direction and its position in the sky. In a second step the shower longitudinal profile is reconstructed. The profile integral, corrected for a small quantity of unseen energy, yields to the cosmic ray energy. 5.1. Reconstruction of the shower axis The reconstruction of the shower axis begins with the determination of the plane containing the shower axis and the fluorescence detector. This plane is called SDP and its reconstruction method is described in [3]. The uncertainty in the SDP depends by the observed angular track length and, of course, by the finite photomultiplier field of view (1.5 ). The uncertainty on the normal to the reconstructed SDP plane is about 0.3 and it has been evaluated using Monte Carlo simulations and analyzing CLF laser shots (whose direction is known). The shower axis position within the SDP is then determined using the measured arrival time of light from the shower in the pixel field view. As discussed in [5], this reconstruction is very poor but it can be significatively improved using the timing information provided by the tanks of the surface array. Using this hybrid reconstruction, the shower axis direction can be estimated with an accuracy of 0.6 and the core location with a resolution of 50 m. 5.2. Reconstruction of the longitudinal profile The reconstruction of the shower longitudinal profile [3] needs a detailed knolwedge of the various physical processes which determine the fluorescence light production, the propagation through the atmosphere and detection by the FD telescopes. One of the most important parts of the reconstruction chain is a detailed knowledge of the emission process of the ultraviolet fluorescence light. This process is studied in several experiments [11]. For the main emission line (337 nm) about 5 photons per MeV of deposited energy are emitted at the pressure of 1000 hp a. Thesystematic uncertainty on the absolute fluorescence yield is currently 13% [12,13]. The fluorescence spectrum and its dependence on atmospheric conditions (air pressure, temperature and relative humidity) should be also known. The total systematic uncertainty introduced by the fluorescence emission process in the reconstruction chain is about 15%. Once the atmospheric medium has been excited, only a small fraction of the fluorescence radiation is emitted in the FD telescope field of view. Being the fluorescence light emitted isotropically, this fraction is proportional to 1/r 2, where r is the distance between the shower track and the telescope. The precise hybrid geometry reconstruction introduces an uncertainty in the profile measurement typically at level of 2%. Not all photons produced in the telescope field of view are detected. During its travel toward the detector the light may be scattered outside the FD field of view by the air molecules and the air suspended aerosol. As previously said in section 4 the Rayleigh scattering with the air molecules is a very well known process. Its determination needs the measurement of the atmospheric profile. The uncertainty in the shower size recon-

42 V. Verzi / Nuclear Physics B (Proc. Suppl.) 165 (2007) 37 44 struction induced by the day-to-day variations of the Malargue monthly model profiles is 0.5% [8]. The light scattering with the aerosol is characterized using the measurements provided by the aerosol monitoring instrumentations. A very conservative estimation of the systematic uncertainty in the shower size introduced by the aerosol measurements is 10%. The photon scattering cross sections with the air molecules and aerosol depend on photon wavelength (λ). Also the FD telescope quantum efficiency (ɛ λ ) has a wavelength dependence. Therefore the detected signal will depend on a convolution over λ of the fluorescence spectrum, scattering cross sections and ɛ λ. The FD quantum efficiency depends mainly on the PMT quantum efficiency. Currently the reconstruction algorithm uses the PMT relative quantum efficiency measured by the manufacture. The absolute value of the FD quantum efficiency is included in the drum calibration constants. The systematic uncertainty on the reconstructed shower size due to the wavelength dependence of the telescopes quantum efficiency was conservatively estimated at the level of 10%. Preliminary measurements of this dependence using the drum multi-wavelength calibrator well agree with the data currently used in the analysis. As the multi-wavelength drum calibrator operate stably, the current 10% systematic uncertainty will be significantly reduced. It has to be noticed that in fact we are dealing on a relative measurement, very different and easier than the absolute calibration. As previously said, the systematic uncertainty on the drum absolute calibration constants is 12%. The laser shots fired by the CLF provide a good tool to cross-check the absolute photometric calibration of the FD telescopes. Fitting the longitudinal profile of vertical tracks, recorded by the FDs under extremely clear atmospheric conditions, is possible to reconstruct the laser power. The laser power is also measured with dedicated energy probes [10] installed at the CLF with an uncertainty of 10%. The consistency in the reconstructed and measured CLF energies is at the level of 15% consistent with the current level of uncertainty in the FD absolute calibrations, laser beam measurement and atmosphere. Events 90 80 70 60 50 40 30 20 10 Roving laser (All runs) Entries 991 2 χ / ndf 32.01 / 31 Constant 70.55 ± 2.81 Mean 1.718 ± 0.179 Sigma 5.445 ± 0.131 0-30 -20-10 0 10 20 30 (E rec -E shot )/E shot (%) Figure 4. An example of distribution of the relative difference between recontructed and measured laser energy. The data sample refers to 337 nm nitrogen laser shots fired in front of several bays of Los Leones and Coihueco telescopes under nearly aerosol-free conditions. A more precise test has been performed using 337 nm nitrogen laser which can be moved in front of the FD telescopes at distances of about 3 km. In comparison to the CLF, the shorter distance reduces the sensitivity to atmospheric effects. Figure 4 shows an example of distribution of the relative difference between reconstructed and measured laser energy obtained for several shots fired in front of different telescopes under nearly aerosol-free conditions. Due to the finite light spot dimension, the camera signal at a given time is spread among several pixels. The signal collection algorithm is based on the maximization of the signal to noise ratio when considering photomultipliers that are geometrically close to the showertrackat the camera. An alternative light collection method, which is based on a realistic model of the spot light distribution at the camera focal surface, has been also implemented. The spot model accounts for

V. Verzi / Nuclear Physics B (Proc. Suppl.) 165 (2007) 37 44 43 the telescope spherical aberration and the finite shower lateral size. The expected spot light distribution is used to define which pixels should be used to build the light profile and the temporal window inside which the PMT signal is selected. A detailed knowledge of the spot light distribution allows to account also for the small camera inhomogeneities due to the presence of the Mercedes. The systematic uncertainty in the reconstructed shower size due to the light collection algorithm is 5%. The relativistic shower charged particles emit Cherenkov radiation in the direction of the shower axis. The small angular spread of this radiation corresponds to the angular spread of the shower particles. This spread has been parametrized using Monte Carlo simulations [14]. In general, Cherenkov photons present at a given point of the shower can be Rayleigh and scattered with aerosol in the field of view of the fluorescence detector. For some specific shower geometries, namely pointing towards the detector, Cherenkov photons can enter directly in the detector field of view. Taking into account the atmospheric transmission and convolving the Cherenkov spectrum with all physics processes which depend on photon wavelength, it is possible to estimate the amount of Cherenkov radiation which is detected for a given shower longitudinal profile. Considering the fluorescence and Cherenkov contribution, it is possible to calculate the shower longitudinal profile from the measured PMT signal, i.e. the energy deposit as a function of the atmospheric depth de/dx(x). The atmospheric depth X [g/cm 2 ] is calculated using the shower geometry and the atmospheric profile. The best estimation of the shower energy and the depth at which one has the maximum profile (X max ) is obtained fitting the data with a Gaisser-Hillas functional form. Figure 5 shows the reconstructed longitudinal profile for a shower fully contained in the FD field of view. The profile integral provides a calorimetric measurement of the cosmic ray energy. In fact only a small correction is made for the unseen energy. This correction is energy dependent and it ranges from 5 to 15% with a systematic uncertainty of about 3% [15]. ] 2 de/dx [MeV/g/cm 60 50 40 30 20 10 9 10 19 E = 3.5 10 ev 2 X max = 724 g/cm 0 200 400 600 800 1000 1200 1400 2 X [g/cm ] Figure 5. Longitudinal profile for an FD event. The dots show the data and the red line is the fitted energy deposit. The total systematic uncertainty on the reconstructed shower energy is 25% and it has been obtained combining all the uncertainties affecting the reconstruction chain. 6. CONCLUSIONS Three of the four fluorescence detectors of the Pierre Auger Observatory operate stably. The fourth detector is under construction and it will be in operation in the first part of 2007. The main detector features and the instruments used for the telescopes absolute calibration and for the atmospheric monitoring have been described. The systematic uncertainties in the various steps of the shower longitudinal profile reconstruction have been discussed. The total systematic uncertainty on the cosmic rays energy has been evaluated to be 25% with an high expectation to reduce it significantly.

44 V. Verzi / Nuclear Physics B (Proc. Suppl.) 165 (2007) 37 44 REFERENCES 1. K.Greisen, Phys. Rev. Lett. 16 (1966) 748. 2. G.T.Zatsepin and V.A.Kuzmin, JETP Lett. 4 (1966) 78. 3. J.Abraham et al., Nucl. Instrum. Meth. A523 (2004) 50. 4. J.Ridky for the Pierre Auger Collaboration, this proceedings. 5. M.Mostafa for the Pierre Auger Collaboration, this proceedings. 6. The Pierre Auger Collaboration, in proceedings 29th ICRC 8 (2005) 5. 7. The Pierre Auger Collaboration, in proceedings 29th ICRC 8 (2005) 101. 8. The Pierre Auger Collaboration, in proceedings 29th ICRC 7 (2005) 123. 9. The Pierre Auger Collaboration, in proceedings 29th ICRC 8 (2005) 347. 10. The Pierre Auger Collaboration, in proceedings 29th ICRC 8 (2005) 335. 11. Contributions to 4th air fluorescence workshop, May 17-20, Prague-Pruhonice. 12. M.Nagano et al, Astropart. Phys. 20 (2003) 293. 13. M.Nagano et al, Astropart. Phys. 22 (2004) 235. 14. F.Nerling et al, in proceedings 29th ICRC 7 (2005) 127. 15. H.Barbosa et al., Astropart. Phys. 22 (2004) 159.