P A M E L A. An Antiproton, Positron Experiment. on a. Polar Orbit Satellite. Technical report

Transcription

1 P A M E L A An Antiproton, Positron Experiment on a Polar Orbit Satellite Technical report July 1996

2 University and INFN, Bari (Italy) M. Ambriola, R. Bellotti, F. Cafagna, M. Castellano, F. Ciacio, M. Circella, G. De Cataldo, C.N. De Marzo, N. Giglietto, B. Marangelli, N. Mirizzi, P. Spinelli Tata Institute of Fundamental Research, Bombay (India) S.A. Stephens University and INFN, Firenze (Italy) O. Adriani, F.M. Brancaccio, G. Castellini, R. D'Alessandro, P. Papini, A. Perego, S. Piccardi, P. Spillantini, V. Vignoli Laboratori Nazionali INFN, Frascati (Italy) S. Bartalucci, G. Basini, M. Ricci NASA Goddard Space Flight Center, Greenbelt (USA) L.M. Barbier, E.R. Christian, J.F. Krizmanic, J.W. Mitchell, J.F. Ormes, D.E. Stilwell, R.E. Streitmatter Particle Astrophysics Laboratory, New Mexico State University, Las Cruces (USA) S.J. Stochaj Moscow Engineering and Physics Institute, Moscow (Russia) A.M. Galper, S.V. Koldashov, M.G. Korotkov, V.V. Mikhailov, A.A. Moiseev, J.V. Ozerov, S.A. Voronov Space Radiation Laboratory, California Institute of Technology, Pasadena (USA) A.J. Davis, R.A. Mewaldt, S.M. Schindler II University and INFN, Roma{Tor Vergata (Italy) R. Cardarelli, M. Casolino, M.P. DePascale, F. Giannini, P. Picozza, A. Morselli, R. Sparvoli Siegen University, Physics Department, Siegen (Germany) M. Hof, W. Menn, M. Simon Royal Institute of Technology, Stockholm (Sweden) D. Bergstrom, P. Carlson, T. Francke, C. Fuglesang, N. Weber University and INFN, Trieste (Italy) G. Barbiellini, M. Boezio, V. Bonvicini, U. Bravar, P. Schiavon, A. Vacchi, N. Zampa

3 Contents I The PAMELA experiment 5 1 Introduction Historical preamble The balloon ight program of the WiZard collaboration, results and perspectives The Russian Italian Mission (RIM) program Small satellite borne experiments The RIM{2 experiment \PAMELA" The RIM{3 experiment \GILDA" The PAMELA experiment scientic objectives Antiprotons Positrons and Electrons Search for antimatter Additional objectives Modulation of galactic cosmic rays (GCR's) in the heliosphere Solar are particle spectra SCR's propagation and acceleration in the internal heliosphere Researches on Earth's magnetosphere and magnetic eld Research of the stationary and disturbed uxes of high energy particles in the Earth radiation belt Anomalous component of cosmic rays Conclusions The PAMELA telescope The PAMELA telescope concept The technical choices and the baseline design The PAMELA rates and data ow The spacecraft Resource{Arktica and the mission prole Spacecraft and Mission description \Resurs{Arktika" SC purpose and tasks On{board instruments composition The informational instrumentation Main systems of the service instrumentation Structure PAMELA instrument design Integration and tests Transportation Background conditions and expected exposition Models used for background calculation Results of evaluations Exposition Numbers of registered particles and expected volume of the stored information Measurements in the radiation belts Expected measurement results in the radiation belt

4 6 Interfaces Power Supplies for the PAMELA devices Introduction Main PS system and interface HV Power Supplies for PM and TRD PS for the Silicon detectors Low voltage PS for the tracker, TRD and calorimeter readout electronics V PS for the sub-detectors and the general processing and control electronics Control of operation mode of PAMELA devices Reception of time code Central processing and Slow Control Unit Telemetry system for the slow control of the apparatus II Detectors in PAMELA 67 7 The scintillation counter hodoscopes Requirements and design goals for the time{of{ight scintillator system Experience in TOF design and fabrication The time{of{ight system General description Photomultipliers Light guides Magnetic shields Electronics Construction Weight Performance PAMELA TOF Summary The anticoincidence counters PAMELA anticoincidence Summary Transition Radiation Detector Introduction Radiator optimization Straw tube detector for transition radiation Cathode materials Anode wires Straw tube electrical connections End plate Tests on straw tubes prototypes with radioactive source Straw chamber tests in vacuum on a beam line TRD design Straw tube TRD expected performances TRD electronics TRD data acquisition and local processing ADC based read{out The magnetic system Magnetic materials General consideration on PAMELA magnet conguration PAMELA Magnet design Magnetic map calculation Test and space qualication Time schedule and plane of the work

5 10 The silicon tracker Experience with Silicon Detectors General requirements and design goals of the silicon tracker Optimization of the geometrical design Energy Range for Antiproton Flux Measurements and for Antihelium Search `State of the art' of silicon microstrip detectors Double sided detectors AC coupled detectors Double metal detectors Double Sided Double Metal AC coupled technology for the PAMELA sensors Spatial resolution and results of the tests Mechanical Design and Construction of PAMELA Tracker Description of the Mechanical Support Design Calculations on the Structure Alignment of the detector Readout electronics Front end Design of the hybrid Analog to Digital Conversion Preprocessing of data with a DSP / Data Compression Sequencer Summary on power consumption Mechanical positioning and geometrical dimensions of the electronic boards Calorimeter The physics task The starting point Balloon ight results CAPRICE The Proposed PAMELA calorimeter MonteCarlo simulation of the PAMELA Calorimeter The Si detector The electronics The mechanics III Trigger, data transmission and general arrangement Trigger logic and second level trigger Trigger signals Trigger signals for the SAA conditions The second level trigger Neural network trigger hardware Trigger organization Test results Data transmission and reception Introduction Structure of the Russian remote setting system Trends of system development Analysis of initial data PAMELA device capacity Block{diagram of the on board information system Ways of arranging data transmission and reception Arrangement of data reception in Russia Data reception in Italy RGS with large antenna RGS with small antenna Check of the PAMELA experiment Conclusion

6 14 The Ground Support Equipment The purpose of GSE complex Structure of GSE GSE composition GSE functioning for various kind of tests Software Constructive features of GSE Operational requirements IV Schedule, work organization and costs The time schedule for the PAMELA experiment Technical requirements Laboratory Prototype Qualication Model Dimensional Model Flight Model Time Schedule The work organization for the PAMELA experiment The cost evaluation of the PAMELA experiment 185 V Appendix 189 A Organization of PAMELA device 191 A.1 Technical requirements to equipment, installed on SC (by VNIIEM) A.1.1 Introduction A.1.2 Technical requirements A.1.3 Requirements on stability, strength and resistance to environment eects A.2 Types of tests A.3 Testing facilities

7 Part I The PAMELA experiment 5

8

9 Chapter 1 Introduction 1.1 Historical preamble In the seventies many experiments devoted to the study of electromagnetic signals arriving from the space were performed on satellites, outside the terrestrial atmosphere. Following the many unexpected results of these experiments and of other ground based observations of the electromagnetic spectrum, the National Aeronautics and Space Administration of USA planned the construction and launch in the following two decades of four large observatories for the permanent observation of dierent ranges of the electromagnetic spectrum: HST dedicated to the visible, SIRTF to the infrared, AXAF to X{rays and GRO to gamma{rays. In the same period, the number of experiments own in space for observing high energy cosmic rays was small; only two signicant experiments were own: the IK{15 experiment of Grigorov et al. in 1968 and the HEAO-3 in The third experiment own until now, the CRN experiment on board of Spacelab{2, was own later, in In 1979 the Committee on Space Astronomy and Astrophysics of the USA Academy of Sciences recommended a systematic study of the Cosmic rays [1] in order to complement the eort made in the study of the electromagnetic spectrum. The basic reason for this recommendation was the awareness that cosmic rays bring to us pieces of matter from astrophysical objects, a variety of about 400 isotopes, each with its own story of synthesis by a star, followed by propagation, acceleration and interaction with the electromagnetic radiation and the interstellar matter. About 80 of the species were within reach of experimental techniques, and could give important information on stars in formation, warm interstellar clouds, interstellar matter and magnetic elds in interstellar matter, stellar winds, remnants in supernovae, and spinning neutron stars, as well as more specic information concerning the cosmic rays sources, connement in the Galaxy and acceleration mechanisms. Further, unexpected results from the measurements of the antiproton [2] and positron [3] content of cosmic rays in experiments own in the late seventies by balloons raised a keen interest among elementary particle physicists and astrophysicists. Fundamental questions concerning the nature of the dark matter and the matter{ antimatter symmetry of the Universe were invoked to explain the abundances of these antiparticles. The lack of conrmation of the predictions of the Grand Unication Theories at their lower level, such as proton decay, the existence of magnetic monopoles and the neutrino masses and oscillations, continue today to fuel this interest. It was emphasized by the Committee that the fundamental questions raised by these rst scanty and sometimes contradictory results required an eort for a systematic and long exposure collection of good quality data. Further, direct observation of cosmic rays in space could stand as a new tool complementing the X{ray, UV, optical, infrared and radio{wave observations of the electromagnetic spectrum. Another USA National Academy report on this subject was published in 1982 [4], and in the same year NASA convened a study group [5] for implementing the recommendations of the two Academy reports. Finally, an organized program for direct observation of cosmic rays [6] was elaborated by the Cosmic Radiation Working Group established by NASA in response to the directive of developing a manned Space Station to be assembled in orbit in the middle 90's. In this program, besides an increased and systematic eort in ballooning activity and an increased number of Shuttle missions, two major projects were suggested: an explorer spacecraft ying outside the magnetosphere devoted to the high resolution study of individual isotopes in the energy range below 1 GeV/nucleon, and a superconducting magnet spectrometer facility for the future Space Station, capable of conducting a wide variety of measurements on the energetic particles above 1 GeV/nucleon. Additionally, studies for the long range future were promoted for an interstellar probe, experiments on polar platforms and large assemblies in space. As a consequence of the Challenger disaster, all opportunities for Shuttle ights and reights were canceled. Also, the USA balloon ight program had a long hiatus, and only a few balloon{borne experiments devoted 7

10 to the antiproton/proton ratio measurement could be own in the late eighties. The Explorer Experiment (Advanced Composition Explorer { ACE {) to measure isotopes will be launched in The superconducting magnet spectrometer elaborated by the CR Working Group was given denition [7] by aninternational denition team appointed by NASA, composed by scientists from the USA, from Italy and from Germany and Japan. The spectrometer consisted of a core facility servicing dierent detection systems installed in its magnetic eld. These detectors could be substituted, improved or complemented, as well as stored when not in use. The volume and strength of the magnetic eld were chosen to match the maximum energies that could be reached by the served experiments with regard of their geometrical acceptance Antiproton/proton ratio Kinetic energy [GeV] Figure 1.1: measurements of the p/p ratio [29]. The curve is the prediction for the secondary production in the interstellar medium [30] The superconducting magnet spectrometer was oered by NASA (with the name of ASTROMAG) as a facility in the rst Announcement of Opportunity for the early payloads [8] for the Freedom Space Station. Among the experiments proposed for ASTROMAG in response to the NASA announcement of opportunity, three were selected for the rst experimental phase: the experiment SCIN-MAGIC for measuring elemental composition and energy spectra from to more than ev/particle; the LISA experiment devoted to the measurement of elemental energy spectra and isotopic composition of cosmic rays; and the WiZard experiment measuring the energy spectra of elementary particles and antiparticles and searching for antinuclei. 8

11 At the end of 1990, the construction of the Freedom Space Station was delayed and descoped, and the attached payload program, including the ASTROMAG facility, was indenitely postponed. A reduced version of ASTROMAG to be put in orbit on a dedicated satellite (ASTROMAG Free{Flyer) was proposed [9]. It was approved and added to the list of the other already planed missions of this class, but could not upset the foreseen launched priority. Positron fraction e + /(e - +e + ) Energy [GeV] Figure 1.2: measurements of the positron fraction in the cosmic rays as function of the energy [31]. The curve is the prediction due to the secondary production using the Modied Leaky Box Model [32]. 1.2 The balloon ight program of the WiZard collaboration, results and perspectives. The WiZard collaboration, which had already begun to prepare prototypes of the instruments and to use them in balloon ights, didn't dissolve and continued its test and balloon ight activities. It also began to look for other possibilities for ying these instruments in experiments that could recover the physics program planned with ASTROMAG. 9

12 The balloon ight program is now a \basic" activity of the collaboration, both for testing instruments and training students, and also for measuring the antiparticle spectra up to the maximum allowed energies by the ballooning technique (about 30 GeV, afterward the measurements are dominated by the atmospheric background). Four balloon ights, in '89, '91, '93 and '94, gave new results in the antiparticle component measurement [10, 11, 12, 13, 14], as well in other components of the cosmic rays [15, 16,17,18, 19, 20, 21]. Particularly valuable are the results obtained for the p/p ux ratio, attaining the 20 GeV energy, and in the e + /(e + +e, ) ux ratio. These results are reported in Fig. 1.1 and Fig. 1.2, together with those obtained by others authors. A few other ights will be performed in the future, the rst one in the spring of '97 from Fort Sumner (USA). The usual duration of balloon ights (about 20 hours of data taking) is too short for an eective hunting of antinuclei. To study this channel a program aiming to the development of suitable high lifting capacitance superpressurized balloons (named the "Stratostation program") was promoted in '92 in collaboration with Alenia Spazio and Winzen Corporation, in order to explore the possibility of maintaining oating at km very heavy telescopes for a very long time. This technique, if suitably developed, could result useful for other measurements (e.g. very high energy gamma rays) and in other elds (astronomic telescopes, Earth observation). 1.3 The Russian Italian Mission (RIM) program. However it is clear to us that the originary Physics program of Astromag can be recovered only by ying magnetic telescopes outside the atmosphere, where both the lack of the atmospheric background and the long duration of the data taking can guarantee the needed sensitivity. After a long series of attempts and proposals [22] submitted to ESA for the M2 mission in 1990, attached payloads to the MIR station, Wizard-sat [23] submitted to Italian Space Agency in '92, CORE [24] submitted to ESA for the M3 mission in 1993) a good opportunity was nally oered by the collaboration with the VNIIEM institute of Moscow, for sending in space cosmic ray telescopes as \piggy{back" in someone of their satellites of the Resurs series. The opportunity is unique both for the very favourable attitude of these satellites (looking to the Earth surface, and indeed out of the Earth on the other side) and for their polar orbit, allowing to cover the whole energy range down to very low energies, highly enriching the possible physics program. Furthermore their orbit is quasi circular at about 700 km, what garantees a very long permanence in ight, and sun{syncronous, what maximizes the electric power availability. The group of Prof. Arkady Galper, of the MEPI of Moscow, acted as interface with the VNIIEM institute, so that in '93 it was elaborated a Russian Italian Mission (RIM) program, of what the RIM{2 experiment, that we named PAMELA, is the central part. The RIM program follows a hinged approach consisting in the several steps below described Small satellite borne experiments The rst step of the RIM program consists in a number of small satellite borne experiments for studying the low energy cosmic ray background and the albedos in the low Earth orbits considered for the study of cosmic rays (MIR orbit and polar orbit). From a technical point of view, these small experiments have been conceived for testing the large area silicon microstrip detectors that will be used in PAMELA and in other possible future space experiments, qualifying them for their use in space, testing their assembly elements, materials and electronics. They work in \stand alone" mode, i.e. the electronics can operate in a self{triggered mode without requiring other sensors to built an external trigger. The experiments RIM{0.1 and RIM{0.2 on board of the MIR space station. The rst experiment (RIM{0.1 experiment, called Si{eye{1 [25]) is taking data on board of the Russian MIR Station from October '95, and a second one (RIM{0.2 experiment, called Si{eye{2) will be brought to MIR at the end of '96. The mass of both these two experiment isvery small, less than 5 kg, and the needed electric power of a few Watts only. For both of them the detector consists of a telescope of 6 large area silicon microstrip sensors for registering, tracking and measuring the energy loss of highly ionizing particles (mainly slow protons and nuclei). They can be mounted on a special helmet on the head of an astronaut for registering coincidences between these particles and the light ash seen by him during the ight. The RIM{1 experiment \NINA". A third experiment (RIM{1 experiment, called NINA [26]) will y in spring '97 as \piggy{back" of a satellite of the same series of that on which the PAMELA experiment is foreseen to y (the above mentioned satellites of the series Resurs built by the VNIIEM institute of Moscow), 10

13 and travelling just in the same polar orbit. Also the mass of the NINA payload (25 kg) and its electric power consumption (40 W) are small. The detector is a telescope of 32 large area silicon microstrip sensors, that will register track and measure the energy loss of protons and nuclei up to about 100 MeV/nucleon. The NINA experiment will indeed supply all the needed information concerning the rates and the albedos that will be of interest in the lowest energy range of the PAMELA experiment The RIM{2 experiment \PAMELA". The second and central step ot the program is the RIM{2 experiment\pamela", devoted to reach the scientic goal of measuring antiproton and positron spectra up to more than 100 GeV, and hunting for antinuclei with a sensitivity two order of magnitude higher than that reachable by balloon borne experiments. These and the other scientic goals will be discussed in the following chapter. The ight is planned for the end of '99 on board of the Russian satellite Resurs-Arktika, built by the above mentioned VNIIEM. Details of the mission and of the instrument are reported in the successive chapters The RIM{3 experiment \GILDA". The PAMELA experiment was from the very beginning conceived as an important and signicant step in the study of the antimatter component of cosmic rays, but not necessarily as the last one, owing the essential limitation of the magnetic eld strength due to the use of a permanent magnet. For this reason all the instrument equipping PAMELA are designed \modular" and \extendable", to orientate them to the possible realization of large detectors for large scale experiments. In particular the construction technique and the consumption/channel of the imaging calorimeter equipping PAMELA are suitable for designing a \standing{alone" instrument that could cover wide areas with high granularity, and could be operated in a self triggering mode for detecting very high energy gamma rays with good angular and energetic resolutions. Such an instrument is the basis of the RIM{3 experiment (called GILDA [27]), that could be a successive experiment own on board of a Resurs series satellite. This instrument has already been assumed as a starting point for other large experiment for the detection of very high energy gamma rays in space. And infact this kind of imaging calorimeter techniques is included in the GLAST proposal [28] and envisaged in the Russian letter of the intent for the detection of very high energy gamma ray on board of the International Alpha Space Station (ISSA). 11

14 12

15 Bibliography [1] Committee on Space Astronomy and Astrophysics: "A strategy for Space Astronomy and Astrophysics for the 1980's", NAS report [2] R.L.Golden et al., Phys. Rev. Letters 43 (1979) 1196; E.A.Bogomolov et al., 16th Int. Cosmic Ray Conf., Kyoto 1979, 1, 330; A.Bungton et al., Ap. J. 248 (1981) [3] J.L.Faneslow et al., Ap. J. 158 (1969) 771; A.Bungton et al., Ap. J. 199 (1975) 699; [4] Astronomy Survey Committee: "Astronomy and Astrophysics for the 1980's", NAS report [5] Cosmic Ray Program Working Group: "Cosmic Ray Program for the 1980's", NASA report, August [6] Cosmic Ray Program Working Group: "The Particle Astrophysics Program for ", NASA report, October [7] Astromag Denition Team: "The Particle Astrophysics Magnet Facility ASTROMAG", NASA report, May [8] Space Station Attached Payloads Announcement of Opportunity, NASA report OSSA 3-88, July [9] ASTROMAG Free{Flyer, NASA GSFC report, May 1991, Vol. 1 and 2 [10] Golden R.L. et al., Ap. J., 436 (1994) 769. [11] Basini G. et al., Proc. XXIV Int. Cosmic Ray Conf., Rome, 3 (1995) 1. [12] Aversa F. et al., Proc. XXIV Int. Cosmic Ray Conf., Rome, 3 (1995) 9. [13] Barbiellini G. et al., Astron. Astrophys., 309 (1996) L15. [14] Hof M. et al., Proc. XXIV Int. Cosmic Ray Conf., Rome, 3 (1995) 60. [15] Bellotti R. et al., Phys. Rev. D, 53 (1996) 35. [16] Brunetti M.T. et al., J. Phys. G, 22 (1996) 145. [17] Grimani C. et al., Proc. XXIV Int. Cosmic Ray Conf., Rome, 4 (1995) [18] Basini G. et al., Proc. XXIV Int. Cosmic Ray Conf., Rome, 1 (1995) 585. [19] M.P.De Pascale et al., J. Geophys Res., 98 (1993) [20] Papini P. et al., Proc. XXIV Int. Cosmic Ray Conf., Rome, 4 (1995) [21] P.Papini et al., Proc. 23rd ICRC (Calgary) 1 (1993) 579; [22] "The EUOROpean Matter Antimatter Space Spectrometer EUROMASS experiment", Proposal in response to the call for proposals for the next medium-size project (M2) D/Sci/RMB/val/3080. [23] The WiZard collaboration: "WiZard-Sat", October [24] "The COsmic Ray Experiment CORE", proposal for a satellite experiment to observe antimatter, gammarays and other topics in cosmic rays, submitted to ESA in rsponse to the call for ideas for the M3 mission D/Sci/RMB/gc/8085, May [25] \Si{Eye on Mir: rst active detector for the study of Light Flashes in space", Proceeding of the 6 European Symposium on Life Science in Space, 17{21 June, Norvay. [26] G. Barbiellini et al., 24th Int. Conf. on Cosmic Rays (Rome), 3 (1995)

16 [27] G. Barbiellini et al., NIM, A354 (1995) 547. G. Barbiellini et al., Nucl. Physics, B43 (1995) 253. [28] Atwood W. et al., NIM, A342 (1994) 302. [29] Golden R.L. et al., Astrophys. Lett., 24 (1984) 75. Bogomolov E.A. et al., Proc. of the 20th Int. Conf. on Cosmic Rays (Moscow), 2 (1987) 72. Orito S. et al., Proc. of the 24th Int. Conf. on Cosmic Rays (Rome), 3 (1995) 76. Labrador A.W. et al., Proc. of the 24th Int. Conf. on Cosmic Rays (Rome), 3 (1995) 64. Bungton A., Shindler S.M. and Pennypacker C.R., Astrophys. Journal, 248 (1981) Hof M. et al., Proc. of the 24th Int. Conf. on Cosmic Rays (Rome), 3 (1995) 60. [30] Gaisser T.K. and Schaer R.K., Astrophys. Journal, 394 (1992) 174. [31] Golden R.L. et al., Astrophys. Journal, 436 (1994) 769. Bungton A. et al., Astrophys. Journal, 199 (1975) 669. Fanselow J.L. et al., Astrophys. Journal, 158 (1969) 771. Muller D. and Tang K., Astrophys. Journal, 312 (1987) 183. Barwick S.W. et al., Phys. Rev. Lett., 75 (1995) 390. Aversa F. et al., Proc. of the 24th Int. Conf. on Cosmic Rays (Rome), 3 (1995) 9. Basini G. et al., Proc. of the 24th Int. Conf. on Cosmic Rays (Rome), 3 (1995) 1. Clem J.M. et al., Proc. of the 24th Int. Conf. on Cosmic Rays (Rome), 3 (1995) 5. Daugherty J.K. et al., Astrophys. Journal, 198 (1975) 493. Golden R.L. et al., Astron. and Astrophys., 188 (1987) Barbiellini G. et al., Astron. Astrophys., 309 (1996) L15. [32] Protheroe R.J., Astrophys. Journal, 254 (1982) , vol. 394, p

17 Chapter 2 The PAMELA experiment scientic objectives The observational objectives of the PAMELA instrument are to measure the spectra of antiprotons, positrons and nuclei in a wide range of energies, to search for primordial antimatter and to study the cosmic ray uxes over half a solar cycle. PAMELA will be able to measure magnetic rigidities (momentum/charge) up to its Maximum Detectable Rigidity (MDR) of 368 GV/c. Data gathered with the PAMELA instrument will deal with a wide range of fundamental issues. These include: 1. the role of Grand Unied Theories in Cosmology in relation to antimatter and dark matter; 2. the understanding of the acceleration and propagation of cosmic rays; 3. the role of solar, terrestrial and heliospheric relationships to energetic particle propagation in the heliosphere. The PAMELA observations will extend the results of balloon{borne experiments over an unexplored range of energies with unprecedented statistics and complement information gathered from Great Space Observatories and ground{based cosmic{ray experiments. These observational objectives can be schematically listed in the following points: measurement of the p spectrum up to 100 GeV (present limit 12 GeV); measurement of the positron spectrum up to more than 100 GeV (present limit 30 GeV); search for antinuclei with a sensitivity of some unity in10,8 in the He/He ratio (present sensitivity limit about 10,5 ); measurement of the p spectrum down in energy to less than 100 MeV; measurement of the positron spectrum down in energy to less than 100 MeV; continuous monitoring of the cosmic rays solar modulation before, during and after the 23rd maximum of the solar activity; study of the time and energy distributions of the energetic particles emitted in solar ares. The last four objectives are in the reach of PAMELA because the satellite travels in a polar orbit, indeed spending a large fraction of its time in the high latitude and polar regions, where the cut{o due to the terrestrial magnetic eld is negligible. The scientic relevance of these four objectives is enhanced by the length of the mission, that is planned to last not less than three years, but could be prolonged for many other years because of the orbit altitude (see below) and the maximization of the electric power due to its sun-syncronism. In the following sub{sections the above objectives are discussed in some detail. 2.1 Antiprotons Antiprotons have been observed in the cosmic rays since 1979 by balloon{borne experiments [1, 2]; prior to these measurements it was generally expected that all primary cosmic rays experienced the same basic history during their acceleration and propagation. Similarly it was assumed that all secondaries were produced and 15

18 stored in the same regions of the Galaxy. These assumptions were quite adequate to explain the observations of secondary Z3 nuclei in cosmic rays (see, e.g.,[3]). The present experimental situation is shown in Fig. 1.1 of the previous chapter. The data at high energies (5{15 GeV) are in conict and do not allow to give rm indication. However, the data from [4] appear higher than expected if the antiprotons are produced by interactions of the cosmic rays with the interstellar medium (ISM).The overabundance of antiprotons has led to speculations of their origin ranging from models where they are produced in shrouded supernovae [5] to annihilation of Majorana fermions created during the Big Bang [6]. Other models explain the excess in term of how cosmic rays may propagate in the Galaxy (for a review, see [7, 8]), or postulate some kind of more \exotic" processes like the evaporation of mini black holes [9, 10] or the annihilation of supersymmetric particles [11, 12, 13] in the galactic halo. These speculations predict dierent energy spectra at high energies [14]. Because of atmospheric backgrounds and limited ight times, balloon{ borne experiments can only measure the energy spectrum of antiprotons up to about 20 GeV. In addition, the sensitivity of the balloon observations is limited by the diculty in eliminating large uxes of atmospheric secondary particles. PAMELA will be able to measure the energy spectrum of antiprotons up to at least 90 GeV. Recent experiments [15, 16], performed with magnetic spectrometers, have investigated the low{energy spectrum of antiprotons in the few hundred MeV range. These measurements are consistent with secondary origin of low energy antiprotons but the large statistic uncertainties require further investigations to clarify if other sources exist. Secondary production involves collision of cosmic rays with the ISM. The cross{sections for these reactions are well known. One key feature is that the antiproton spectrum produced by such collisions sharply falling below 1 GeV due to kinematics of antiproton production. This is often referred as the \kinematic threshold" for antiproton production. Observation of the shape of the spectrum just above the kinematic threshold is especially interesting because this threshold is the only strongly energy dependent production process known in cosmic rays. By measuring the spectral shape it is possible to measure post{production acceleration or deceleration (e.g. Fermi acceleration) of the particles. Somewhere between 1.2 GV/c and 5 GV/c the p/p ratio should have acharacteristic rise. The shape and position of the rise tells us whether or not there is acceleration of the antiprotons in the ISM after they are produced and, indeed, if they are produced by collisions in the ISM. Results of the behaviour of p spectrum in this region are expected from the data analysis of the balloon{borne experiment CAPRICE own in summer 1994 from Canada. PAMELA's orbit inclination (98 degrees) will allow to perform a high{statistics measurement ofantiproton spectrum at low energies. 2.2 Positrons and Electrons Positrons and electrons are unique among cosmic rays because they are the lightest charged leptons. Because of their low mass, high-energy electrons and positrons undergo interactions with the ISM which result in severe energy losses at high energies. While most of the observed electrons are believed to be of primary origin, the origin of positrons is yet to be established. Positrons are even harder to observe than antiprotons due to the high ux of protons (about 1000 times higher). All observations to date have suered from the risk of subtracting signicant background. The results are shown in Fig. 1.2 of the previous chapter; the majority of data shows an excess of positrons above the ux expected by the simple leaky-box model and may even indicate a rise in the positron/electron ratio at energies greater than 15 GeV. The positron spectral index below 15 GeV is found to be about -3.1: this value is what one would expect for secondary production without invoking rigidity dependent storage times or radiative energy losses expected at these energies. The spectrum should be steepening in this region but indications are that it may be actually attening. Clearly the positron spectra are not understandable in terms of our conventional model for cosmic ray production and propagation. These direct observations, combined with the observation of a positron annihilation emission line from the galactic disk and the high antiproton uxes give rise to questions such as: 1. are there positrons in the cosmic rays that are not produced as secondaries? 2. is there a relationship between the antiproton and positron excesses? 3. if positrons are indeed all secondary particles, at what point do the radiative losses become important? Observation of positron over a very large energy range should yield new insights into galactic processes. As for antiprotons, PAMELA will aim to measure accurately the spectra of positrons from low (cuto) energies up to the highest energies attainable (at least 100 GeV). Furthermore, together with the measurement of the spectra of electrons, their spectral shapes provide information on the 1. acceleration of e, and the distribution of acceleration sites, 2. cosmic-ray lifetime, and the physical conditions in the containment volume, 3. magnitude of re-acceleration by interstellar shock waves. 16

19 2.3 Search for antimatter Detection of antimatter of primary origin in cosmic rays would be a discovery of fundamental signicance. Cosmic{ray searches that have been made so far have yielded only upper limits of one part in 10 4 for heavy nuclei (Z>2) and one part in 10 5 for helium [17, 18]. Antiprotons and positrons are not direct indicators for the existence of primordial. This is because they are produced by collisions of the cosmic rays with the ISM. The detection of antinuclei in cosmic rays would provide direct evidence of the existence of antimatter in the universe [24]. Three conditions are necessary to generate a matter/antimatter asymmetry from an original Hot Big Bang: baryon nonconservation, CP violation and a non-equilibrium environment [25, 26]. The CP violation is seen in the laboratory, and the rapid expansion of the Universe, following the Big Bang, could easily generate the non{equilibrium environment. Further it is believed that most of the baryons produced in the Big Bang have in fact disappeared. Baryons and photons would be produced in the Big Bang in equal amount but from observation of the 2.7 K cosmic background radiation and the present matter density of the universe we know that only about one baryon remains for ever 10 9 photons (see, e.g. [27]). Because this value is close to the level of CP violation, the current theory suggests that the remaining matter is the remnant of the almost complete annihilation of matter and antimatter at some early epoch, which stopped only when there was no more matter to annihilate. On the basis of {ray observations and other considerations the coexistence of condensed matter and antimatter on scales smaller than that of clusters of galaxies has been virtually ruled out. However, no observations presently exclude the possibility that the domain size for establishing the sign of CP violation is as large as a cluster or supercluster of galaxies. For example, there could be equality in the number of superclusters and antisuperclusters. Similarly, there is nothing which excludes the possibility that a small fraction of the cosmic rays observed at Earth reach our galaxy from nearby superclusters. No quantitative assessment of the probabilities is available, however. Observations of the diuse {ray ux suggest that either there is no antimatter in the intergalactic medium or the intergalactic medium is very rareed [24]. This is not in contradiction with the hypotesis of well separated domains of matter and antimatter. If there is primordial antimatter, antihelium is the most likely form to be detected in cosmic rays due to the fact that in cosmological predictions of the primordial composition of the universe and in stellar composition, helium is the next most abundant element to hydrogen. PAMELA will search for antinuclei with a sensitivity on the He/He ratio better than 10, Additional objectives The continuous determination of the direction, latitude and longitude of the primary cosmic ray electrons, positrons, protons, antiprotons and light nuclei during several years and over a large energy range (90 MeV up to several tens GeV) will provide a great opportunity to investigate other scientic issues that the PAMELA experiment will be able to address as additional objectives besides the primary ones above described. Indeed, during its orbiting around the Earth, the satellite will encounter all that kind of events that are related to the solar activity and to the terrestrial geomagnetic eects. The integration of all data gathered over its mission will provide a signicant complement to the measurements performed so far with dedicated experiments of dierent concepts. The additional objectives of PAMELA are the following: 1. modulation of galactic cosmic rays in the heliosphere; 2. solar are particle spectra; 3. distribution and acceleration of solar cosmic rays (SCR's) in the internal heliosphere; 4. magnetosphere and magnetic eld of the Earth; 5. stationary and disturbed uxes of high energy particles in the Earth's magnetosphere 6. anomalous component of cosmic rays Modulation of galactic cosmic rays (GCR's) in the heliosphere. The research on GCR modulation in the heliosphere in correlation with the solar activity presents a high scientic interest for several aspects. The possibility of studying the energy, charge (and mass), spatial and temporal distributions of cosmic rays in the interstellar space in the vicinity of the solar system can provide a better understanding of astrophysical mechanisms like, for instance, the acceleration of charged particles in a shocked plasma. Galactic cosmic rays, in turn, can be used as a unique tool for deeply understanding the characteristics and the dynamics of the heliosphere. It is not improper to speak of \heliosphere tomography" with cosmic rays. The rst observations of the modulation of cosmic rays dates to the thirties [28]. Later on, 17

20 the mechanism and the process of modulation was understood as determined by the dipole magnetic eld of the Sun [29]. At the beginning of the fties the eects of the periodic variations (11-year, 26-day, 1-day) and of magnetic storms were discovered [30]. Today, the heliosphere is known to be a volume around the Sun with a radius of about 100 astronomical units (AU, 1 AU= cm, or the mean Sun{Earth distance); in the rst AU a supersonic ux of solar wind (a stream of plasma coming from the expanding solar corona) carries the heliospheric magnetic eld (HMF); the eld has a general pattern of an Archimedean spiral centered on the Sun, with many irregularities superimposed on this pattern, and is divided into hemispheres of opposite polarity that form a neutral sheet that originates in the coronal magnetic eld. Between 60 and 100 AU the heliosphere solar wind turns to subsonic velocity due to the resistance of the interstellar gas for what is called the heliospheric termination shock. Eventually, at distances beyond 100 AU, the solar wind is stopped in a region called heliopause which separates the interstellar gas ow from the solar wind itself [31, 32]. A schematic representation of the heliosphere is illustrated in Fig. 2.1 (taken from [32]). Figure 2.1: A schematic representation of the heliosphere (taken from [32]). Obviously, the theoretical description of the GCR's modulation process by the heliosphere is complex. It includes the diuse scattering process on nonuniformities, drift in a magnetic eld, adiabatic energy changes, convective escape, acceleration and deceleration at the passage of shock fronts [32, 33, 34, 35, 36]. Many parameters of the theoretical description remain still unknown mainly because the modulation process is aected by the strong inuence of aperiodic solar activity (Forbush decrease) [37]. The results of the experimental researches outside the atmosphere can provide a signicant contribution to the understanding of the solar modulation problem. For instance, it is enough to mention the direct measurements performed on board the space vehicles Pioneer 10 and 11 [38] and Voyager 1 and 2 [39], Ulysses [40, 41]. The PAMELA mission will carry out its observations before and after the maximum of the 23rd solar activity cycle. The modulation parameters of electrons, positrons, protons and nuclei during this period will be investigated in order to nd any dependence on charge sign, energy (rigidity) and mass of the particles. Latitude and longitude distributions of the observed uxes will be analysed as a function of the solar activity to look for possible correlations. 18

21 2.4.2 Solar are particle spectra Most of the satellite measurements on the composition and spectrum of solar energetic particles from the ares are limited to energies below a few tens of MeV. Experiments carried by GOES{7 and SAMPEX could determine the spectrum of heavy nuclei to about 100 MeV/n for a few ares. Using PAMELA it is possible to measure the energy spectrum of nuclei from helium to at least oxygen from about 100 MeV/n up to an energy where the solar energetic particles can be distinguished from the galactic cosmic rays. Solar ares with generation of similar particles are rare but, as of 1999, when the solar activity approaches its maximum, about ten of such events per year will be available [42, 43, 44]. The composition and the spectral shape of nucleon components could vary from are to are and it would be very exciting to relate these variations to the physical conditions in the are sites. In the case of protons, direct measurements exists up to a few hundred MeV (see e.g. [45]). Indirect evidence from the ground level enhancement of neutron monitor rates shows that in large ares the protons are accelerated up to energies beyond 10 GeV [46]. Direct measurement of the spectral shape of these protons up to this high energy has not been made so far. The present understading is that these protons are accelerated rst during the ash phase, when they are trapped in closed magnetic structures deep inside the atmophere. They might be later accelerated in the corona by shock waves. It is very essential to obtain the exact spectral shape of the proton spectrum to examine the type of acceleration which is in operation. It may also be noted that the total intensity of energetic protons of energy >30 MeV varies by more than 5 orders of magnitude and PAMELA would be in a position to cover a wide range of these. No measurement of electron spectrum exists except for the MEH experiment in the ISEE{3 satellite, by which the electron spectrum is obtained up to about 100 MeV. It is not clear at this stage whether electrons are not accelerated to high energies due to their energy loss mechanisms or that the experiments so far conducted could not detect them. As far as we know that galactic cosmic ray electron spectrum has not been measured outside the atmosphere so far and no experiment has been designed to look for them. From the observed radio, X-ray and gamma ray emission, it is clear that electrons are accelerated at least up to a few tens of MeV. PAMELA will be able to measure the electron spectrum and its relation to the proton spectrum would through new insight into the acceleration mechanism. PAMELA can also look for positrons in solar ares. It was also noticed that in the impulsive ares there is an enhancement of 3 He abundance and in such ares heavy ion enhancement is also noticed [47]. 3 He enhancement sometime exceeds 4 orders of magnitude in such ares. The present understading of the mechanism of 3 He enhancement is the selective acceleration of these ions by cyclotron resonance and the subsequent heating to energies of a few MeV/nucleon. If the impulsive phase is followed by coronal mass ejection, these ions could be accelerated to much higher energies. It is possible to identify the 3 He and D ux variations by PAMELA in a limited energy region and observation provides a valuable information to understand accelerations associated with ares SCR's propagation and acceleration in the internal heliosphere. For the rst time, there will be a opportunity to observe at the same time high solar energy particles with dierent charges and dierent masses. The observation of such particles will not only allow to choose between dierent SCR's acceleration processes on the Sun, but also to gather information on the distribution and acceleration processes of the particles in the internal part of the heliosphere. As in the case of GCR's modulation, the observation of SCR's, the determination of the energy and of the temporal distributions of various components will allow to carry out the internal tomography of the heliosphere: the shock wave propagation and the uxes of solar wind drivers (coronal mass ejections) [48] will be investigated. Moreover, these studies mayhave a signicant impact on better understanding the strong magnetic disturbances in the Earth magnetosphere Researches on Earth's magnetosphere and magnetic eld The experimental data obtained by PAMELA both in conditions of quiet and disturbed Earth's magnetosphere will allow to study the variations of current systems and their inuence on the trajectory of cosmic rays and on the geomagnetic rigidity thresholds [49] Research of the stationary and disturbed uxes of high energy particles in the Earth radiation belt The PAMELA orbit is about 700 km. In this case, for almost one third of its orbits, the instrument will cross the internal part of the Earth radiation belt (the so-called Brazilian anomaly region) giving the opportunity of observing both primary cosmic rays and the high energy particles of the Earth radiation belt itself. Three aspects of these measurements are of special interest: 19

22 1. Until recently it was assumed that high energy particles of the radiation belt were only protons, produced by the decay of albedo neutrons from the atmosphere. Electrons from neutron decay can have energies not higher than several MeV [50]. In the 80's electrons with energy of 20{200 MeV [51, 52] were found in the radiation belt. Moreover, electrons and positrons together with energies up to several hundred MeV have been observed. Separate measurements of the uxes of electrons and positrons in this energy range could give a signicant improvement in conrming the existence of eective processes of particle acceleration in the magnetosphere or in nding evidence for other generation processes. 2. From general reasons it follows that antiprotons can be produced in the radiation belt as a result of the decay of albedo antineutrons [53]. The upper limit on the antiproton to proton ratio that can be achieved by PAMELA is 10,5 {10,6 at energies of some hundreds MeV; this can be considered sucient for antiproton observations. Similarly, if high energy helium nuclei are trapped in the radiation belt, they will be observed by the PAMELA instrument as well. 3. The study of temporal and spatial uctuation of proton and electron uxes will be carried out. Fluctuations can concern the eects on the radiation belt due to solar and magnetic activity, or to Earth activity like, for example, seismic activity and formation of typhoons. It is not excluded that the observation of these uctuations and the study of possible correlations could allow to develop a way to forecast some natural hazardous phenomena on Earth Anomalous component of cosmic rays The anomalous component of cosmic rays consists of partially (or single) ionized interstellar neutral atoms accelerated at the termination shock and penetrating inside the heliosphere. Their energy can vary from 10 up to several hundred MeV which is sucient to reach the vicinity of the Earth and to be observed by satellites at high latitudes or outside the magnetosphere [54, 55, 56]. On the other hand, the atoms of the anomalous component can be trapped by the Earth's magnetic eld in a radiation belt where they are completely ionized [57, 58,59]. Due to the inclination of the PAMELA's satellite orbit (98 degrees) there is an opportunity to carry out the measurement of the anomalous component of cosmic rays both at high latitudes and in the radiation belt of the Earth. 2.5 Conclusions PAMELA is a powerful instrument that will carry out astrophysical observations at a level of sensitivity never reached either by previous experiments or any conceivable high altitude balloon-borne experiment which are limited by the eects of the overlying atmosphere and limited statistics. PAMELA will surpass all available measurements in accuracy and energy range. In the case of particles such as positrons and antiprotons, spectra will be determined over 2 decades of energy to test many predictions. Limits on antimatter/matter asymmetry will be set by a very sensitive search for antihelium in the cosmic rays. Spectra of primary components of cosmic rays will be measured over 2{3 decades in energy with sucient accuracy to look for structures in the spectrum. The availability of long duration exposure provides the rst opportunity for a detailed study of temporal variations of cosmic rays at relativistic energies; the possibility of measuring the spectra of cosmic rays with the same instrument over a signicant portion of a solar cycle from its beginning to beyond its maximum is of major importance. 20

23 Bibliography [1] R.L.Golden et al., Phys. Rev. Lett. 43 (1979) 1196 [2] E.A.Bogolomov et al., Proc. XVI Int. Cosmic Ray Conf., Kyoto, 1 (1979) 330 and Moscow, 2 (1987) 72 [3] J.F.Ormes and P.Frier, Ap. J., 22 (1978) 471 [4] R.L.Golden at al., Ap. Lett., 24 (1984) 75 [5] S.A.Stephens and B.G.Mauger, Ap.Sp.Sci., 110, (1985) 337 [6] S.Rudaz and F.W.Stecker, Ap.J., 325 (1988) 16 [7] S.A.Stephens and R.L.Golden, Sp.Sci.Rev. 46 (1987) 31 [8] G.Basini et al., Riv.Nu.Cimento, vol.12, n.4 (1989) [9] P.Kiraly et al., Nature 293 (1981) 120 [10] S.A.Stephens, Proc. XVII Int. Cosmic Ray Conf., Paris, 13 (1981) 89 [11] J.Silk and M.Srednicki, Phys. Rev. Lett., 53 (1984) 624 [12] F.W.Stecker et al., Phys. Rev. Lett., 55 (1985) 2622 [13] A.Bottino et al., Astropart. Phys., 3 (1995) 77 [14] S.A.Stephens and R.L.Golden: Astron. Astrophys. 202 (1988) 1 [15] Orito S. et al., Proc. of the 24th Int. Conf. on Cosmic Rays (Rome), 3 (1995) 76. [16] Labrador A.W. et al., Proc. of the 24th Int. Conf. on Cosmic Rays (Rome), 3 (1995) 64. [17] G.F.Smoot et al., Phys. Rev. Lett., 35 (1975) 258 [18] Ormes J.F. et al., Proc. Int. Conf. on Cosmic Rays, Rome, 3 (1995) 92. [19] A.Bungton et al., Ap.J., 199 (1975) 699 [20] J.K. Daugherty et al., Ap.J., 198, (1975) 493 [21] J.L.Fanselow et al., Ap.J., 158 (1969) 711 [22] D.Muller and J.Tang, Proc. XIX Int. Cosmic Ray Conf., La Jolla, 2 (1985) 378 [23] R.L.Golden et al, Ap.J., 436 (1994) 769 [24] G.Steigman, Ann.Rev.Astron.Astrophys., 14 (1976) 399 [25] S.Weinberg, Phys.Rev.Lett., 42 (1979) 850 [26] A.H.Guth, Phys. Rev., D23 (1981) 347 [27] V.M.Chechetkin et al., Riv. Nu. Cimento, vol. 5, n.10 (1982) [28] A. Compton et al., Rev. Sc. Instr., V5, p.415,1934 [29] L. Janossy, Zs. Phys., V104, p.430,

24 [30] S.E. Forbush, J. Geophys. Res., V59, p.525, 1954 [31] L.I. Dorman, 12th ECRS, p.21, 1991 [32] M.S. Potgieter, XXIII ICRC, Calgary '93 Invited, Rapporteur, and Highlight Papers p.213 [33] E.N. Parker, Phys. Rev., V110, p.1445, 1958 [34] N. Iucci, 20th ICRC, V8, p.170, 1987 [35] J.R. Jokipii, Adv. Sp. Res., V9, p.105, 1989 [36] J.A. Simpson, Ann. N. J. Acad. Sc., V655, p.95, 1992 [37] Webber et al., Geophys. Res. Lett., V8a, p.6574, 1984 [38] W. Fillius et al., 22nd ICRC, V3, p.395, 1991 [39] L.F. Burlaga and N.F. Ness, J. Geophys. Res., V99, N A10, p.19341, 1994 [40] D. Winterhalter et al., J. Geophys. Res., V99, N A12, p.23372, 1994 [41] S.J. Tappin et al., Astronom & Astrophys., V292, N 1, p.311, 1994 [42] E.O. Fluckiger, 12nd ECRC, p.1, 1991 [43] N. Mandzhvidze, XXIII ICRC, Calgary '93 Invited, Rapporteur, and Highlight Papers, p.157 [44] M.A. Lee, 22nd ICRC, V5, p.293, 1991 [45] Zhang L. et al., XXIII ICRC, Calgary, 3 (1993) 37. [46] Yoshimori M., Soace Sci. Rev., 51 (1989) 85. [47] Reames D.V., Ap. J. Supp.,73 (1990) 235. [48] W. Droge, XXIII ICRC, Calgary '93 Invited, Rapporteur, and Highlight Papers, p.185 [49] L.I. Dorman, Cosmic ray variation and space exploration, North Holland Publ. Co., Amsterdam, V2,1975 [50] W.N. Hess, The radiation belt and magnetosphere, Blaisdell Publ. Co., London, 1970 [51] A. Galper et al., JEPT Lett., V58, N 3, p.497, 1983 [52] S. Voronov et al., JEPT Lett., V43, N5, p.240, 1986 [53] S. Voronov et al., Cosmic Res., V28, N 4, p.603, 1990 [54] A. Morgo-Compero and J.A. Simpson, Phys Rev. Lett., V25, p.1631, 1970 [55] J.H. Chan and P.B. Price, Phys. Rev. Lett., V35, p.539, 1975 [56] J.R. Jokipii et al., COSPAR, Oxford, 1990 [57] N.L. Grigorov et al., INP MSU, Moscow, Preprint N 48/69, 1988 [58] N.L. Grigorov et al., GRL, V18, p.1959, 1991 [59] J.R. Cummings et al., GRL, V20 N 18, p.2003,

25 Chapter 3 The PAMELA telescope 3.1 The PAMELA telescope concept The concept of the PAMELA telescope is the same on which the proposed WiZard experiment was based: a magnetic spectrometer capable of determining the sign of the electric charge with a very high condence degree, and of measuring the momentum of the particle up to the highest energies for which a useful ux of rare particles (as antiprotons and positrons) can be collected; an imaging calorimeter that can give, besides the measurement of the energy released by the interacting particle (and indeed of the extra energy released in an annihilation event), the pattern of the interaction of the particle inside the calorimeter. This last, observing the annihilation pattern of antiprotons and possible antinuclei, allows conrmation of their identity on an\event by event" basis; a precise time of ight counter system; a velocity measurement system to help the calorimeter in the identication of the nature of the particle. The PAMELA experiment can make use of all the experience accumulated in the last years by its proponents, collecting in one experiment an advanced version of the detectors developed for the previous balloon ights, allowing a high selectivity and redundancy for many of the physics goals to which PAMELA may contribute. Most of the proponents of PAMELA contributed from the very beginning to the denition of the ASTRO- MAG facility and to the ongoing WiZard program. They have developed a number of new detectors that were used in the above mentioned balloon ights, and represent the \state of art" in particle detection techniques. In particular two versions of \imaging calorimeters" were developed, one constituted of streamer tubes own in the MASS [1, 2] and MASS2 [2] ights in '89 and '91, and the second one constituted of microstrips implanted in silicon crystals, that was own in two dierent congurations in the TS93 ight in '93 [2] and in the CAPRICE ight in '94 [2]. Two other new instruments, a Transition Radiation Detector (TRD) [3] and a Ring Imaging Cherenkov (RICH) [4] counter were also own in these two last ights. Finally high performance time{of{ight (TOF) systems, in whose development were involved members of the NASA/GSFC and University of Siegen research groups proposing PAMELA, were realized for several balloon-borne cosmic-ray experiments (LEAP [5], SMILI [6], IMAX [7], TIGER [8], ISOMAX [9], and all of the TOF systems for the various WiZard balloon ights). Furthermore, the mature experience gained in previous studies of the antimatter component in the cosmic radiation and of the light isotopes will be of great valuable in determining the right approach to dimensioning the device, choosing the trigger selection and background rejection techniques, handling the collected data and extracting from them the promised physics results. 3.2 The technical choices and the baseline design. The technical choices had to be made taking in account the mass and power limits of the payload, dictated by the launch opportunity, but also protting from technical progress made since the ASTROMAG study. In particular the severe limitations in mass and volume of the payload makes the use of superconducting coils cooled by liquid helium, as was proposed for the ASTROMAG facility, not competitive for producing the magnetic eld of the spectrometer in comparison with a solution making use of a permanent magnet. This is in light of recent progress in the quality and intensity of the magnetic eld that can be obtained by permanent magnets. In fact the life{time of a liquid helium dewar strongly depends by its surface/volume ratio, and can reach several years only for very large dewars, more than 1000 liters in volume. 23

26 10 4 antiproton counts E K [GeV] Figure 3.1: antiproton counts as expected in the PAMELA experiment in one year and with a total eciency of 60%. In the gure are reported the data in the case of secondary production only, using the Modied Leaky Box Model and the Closed Galaxy Model. The expected proton spillover background is reported too. The loss in the magnetic eld intensity due to use of a permanent magnet can be compensated with the dramatic progress made in the recent years in particle localization by the wide application of the silicon microstrip detectors. Some of the proponents of PAMELA have acquired experience in the construction and operation of the silicon vertex detector SMD for the L3 experiment at the LEP of CERN, and introduced these techniques into the collaboration, where extensive experience with silicon detectors and the associated electronics already existed after the construction and operation of the imaging silicon calorimeters for the above mentioned balloon ights. The tests made in '94 and '95 of prototypes of the foreseen sensors for the PAMELA tracker guarantee the successful handling of this dicult technique. The geometrical dimensions of the permanent magnet in the mass limit allocated to the experiment impose the choice between two conicting requirements: 1. the maximization of the bending power, expressed by the Maximum Detectable Rigidity (MDR) parameter; 2. the maximization of the geometrical acceptance, expressed by the Geometry Factor (GF) parameter. The rst choice maximizes the maximum energy reachable in the measurement of the antiproton and positron energy spectra, while the second one maximizes the sensitivity reachable in the search of antinuclei. We chose to maximize the bending power of the magnetic spectrometer, provided that the foreseen rates for antiprotons and positrons at the highest reachable energies are enough high to obtain a useful measurement of their ux. The chosen cross section of the magnet gap is 1416 cm 2, matching the optimal use of the 4 inch silicon wafers to produce large area silicon microstrip sensors. For maximizing the MDR we chose to have a relatively weak eld (0.35 T) on a relatively long magnetic path (44 cm, nearly the maximum allowed by the geometrical constraints to the experiment). The corresponding MDR is 370 GV/c. The magnetic gap and the magnet 24

27 length determine the angular acceptance of the telescope, , to which corresponds the geometrical acceptance of 24.5 cm 2 sr (see Fig. 9.1). The momentum resolution for protons as a function of their momentum is reported in Fig At low momenta it is dominated by the multiple scattering of the particle in the silicon sensors and in their supporting structures, afterward it increases linearly with the momentum. The antiproton counting rate in the given GF is reported in Fig. 3.1 in two hypotheses: 1) only secondary produced antiprotons according to the Modied Leaky Box Model of the Galaxy (lower curve); 2) only secondariy produced antiprotons according to the closed Galaxy Model (CGM) of the Galaxy. Any primary production mechanism will increase these rates. The counting rate foreseen for the positrons is reported in Fig positron counts E K [GeV] Figure 3.2: positron counts as expected in the PAMELA experiment in one year and with a total eciency of 60%. In the gure are reported the data in the case of secondary production only, using the Modied Leaky Box Model. The expected electron spillover background is reported too. In the gures are also reported the expected rates of the spillover protons in the antiproton side and of the spillover electrons in the positron side for the above given MDR. The spillover particles set the measurable range for the antiproton ux to 100 GeV for the MLBM hypothesis (150 GeV for the CGM hypothesis), and to 150 GeV for positrons. At these energies the rates are enough high for a signicatively high statistics measurement on a period of 3 years. The technical choices for the other detectors (besides the obliged choice of silicon microstrip sensors for the tracker) are the following: silicon microstrip sensors for the imaging calorimeter, interleaved with tungsten plates as assorber; this choice minimizes the volume of the calorimeter, and maximizes indeed his geometrical acceptance; the high granularity assumed in PAMELA allows a very good separation between electromagnetic showers and interacting or not interacting hadrons (see later chapter 11); the chosen depth allows a good resolution in the measurement of the energy of electromagnetic particles, further extending the energy spectrum measurements of electrons and positrons; 25

28 Figure 3.3: the PAMELA telescope in the bending view. 26

29 Figure 3.4: the PAMELA telescope in the non{bending view. 27

30 e - /p - ratio momentum [GeV/c] Figure 3.5: Detector rejection powers, taking into account the expected performances of the various sub{detectors described in the Part II, compared with the expected cosmic ray ratios of electron/antiproton from the Modied Leaky Box Model (MLBM) and the Closed Galaxy Model (CGM). The MLBM curve of the e, /p, ratio represents the maximum expected electron background in the antiproton measurement. The PAMELA apparatus is enable to well discriminate the antiproton component in all the energy range. several (5) scintillation counter hodoscopes for the construction of the triggers and for the TOF measurements; the use of several hodoscope allows independent TOF measurements, improving the precision and the safety; the hodoscope structure maximizes the yield of the collected light and the safety of the system; both the hodoscopization and the number of hodoscopes allow a good exibility in constructing the rst level triggers. a Transition Radiation Detector (TRD) for selecting electromagnetic particles from hadrons up to very high energy (1000 GeV); the TRD is based on small diameter straw tubes arranged in double layer planes interleaved by carbon ber radiator. The use of the straw tubes allows to perform many energy loss measurements along the particle trajectory (what brings down to less than 1 GeV the selection capability of the instrument), and to track all particles before their entrance in the magnetic spectrometer, cleaning the sample of the particles accepted at its entrance; a sixth scintillator layer will be located below the calorimeter as a penetration detector; nally a set of scintillation counters covering the top edge and the sides of the magnetic spectrometer completes the telescope, for a further labelling of contaminated events. The PAMELA telescope (see Fig. 3.3 and Fig. 3.3) consists indeed of the following subdetectors (from the top to the bottom): 28

31 two scintillation counter hodoscopes (S11 and S12); the TRD counter, occupying 29 cm in height; one scintillation counter hodoscope (S13); a scintillation counter hodoscope (CAT) on the top of the magnet, labelling the particles not entering the magnet gap; the magnet + tracker system, consisting of 5 permanent magnets, each 8 cm high, interleaving 6 detection planes of the silicon microstrip tracker, each 0.8 cm high, the whole closed in a ferromagnetic screen, 0.2 cm thick, and surrounded on its sides by a system of scintillation counters (CAS) labelling the particles entering the spectrometer from the sides; two scintillation counter hodoscopes (S21 and S22); the imaging calorimeter, consisting of 23 planes of silicon microstrip sensors, interleaved by 23 tungsten sheets, whose physical occupancy in height is 19 cm; the penetration detector (scintillation counter S3). The total height of the spectrometer is cm p + /e + ratio momentum [GeV/c] Figure 3.6: Detector rejection powers, taking into account the expected performances of the various sub{detectors described in the Part II, compared with the expected cosmic ray ratio of proton/positron from the Modied Leaky Box Model (MLBM). The MLBM curve of the p + /e + ratio represents the maximum expected proton background in the positron measurement. The PAMELA apparatus is enable to well discriminate the positron component in all the energy range. 29

32 The side dimensions of the sub{detectors have tobechosen in order to fully cover the geometrical acceptance of the magnetic spectrometer. This is the main reason for locating the imaging calorimeter, which has the highest mass/volume ratio, as near as possible to the magnetic system; consequently the TRD has to be located on the top of the magnetic spectrometer. This has the drawback ofintroducing some extra material on the particle trajectory, before the determination of its momentum by the tracker in the magnetic eld. We kept the criterion that the total material thickness introduced by the TRD should be less, both in radiation and in interaction lengths, of that of a single scintillation counter hodoscope. The imaging calorimeter is, besides the tracker, the most complex detector in the telescope. The silicon microstrip sensors are identical to those own in the last two balloon ights of the WiZard collaboration, TS93 and CAPRICE. The high granularity (3.6 mm in both the transverse coordinates, with a 0.7 X 0 longitudinal pitch) assures a good quality of the pattern of the interaction of the incoming particle with the material of the calorimeter. This feature allows to combine in the same instrument apowerful particle identication capability with a good resolution measurement of the released energy. For the particle identication the criterion is that of the topological dierence between electromagnetic showers and hadronic interactions, allowing to reach contaminations of the two categories of the order of 10,4. Furthermore the interaction pattern becomes very important when an antiproton annihilates inside the calorimetric volume: the detection of the extra energy released in the annihilation, joined to its topological balance, supplies a high condence antiproton identication on an \event byevent" basis up to several GeV. This \event byevent" identication criterion is of the maximum importance for the identication of the interaction of a possible antinucleus. Finally, the total depth of 16 X 0 assures a good containement of the electromagnetic shower iniziated by electrons and positrons, measuring their energy better than in the magnetic spectrometer at all medium and high energies (E>10 GeV). The PAMELA rejection power of electrons and protons in the antiproton and positron measurements are shown in Fig. 3.5 and 3.6. The calorimeter performances ensure a good rejection of electron and proton background and the TRD increases the PAMELA particle discrimination capability of one or two order of magnitude. In Table 3.1 are summarized dimensions, masses, number of readout channels and power consumption of all the elements of the PAMELA telescope. Volume(xyz) Mass R.O.ch Power (cmcmcm) (kg) (#) (W) Scintillation counters Trans. Rad. detector , Anticoincidences (2161) (9301)15 (8301)5 Silicon tracker ( ) , Imaging calorimeter , Fast trigger 5 10 General electronics Power supply Permanent magnet [(23248)5+ Fe shielding] General mechanical 25 structure Total , Table 3.1: Main parameters of the elements of the PAMELA telescope Details and expected performance of all the sub{detectors are given in the part II of the report. The trigger logic, the general arrangement and the mechanical structure supporting these elements, and the general electric and electronic systems of the experiment are reported in the part III. 30

33 Figure 3.7: geomagnetic transmission factor averaged on orbits with dierent inclination respect to the equator for Z=1 particles. 3.3 The PAMELA rates and data ow The PAMELA experiment will be put in polar orbit; therefore the total number of collected events will be high in spite of the small GF. In fact the magnetic cuto due to the Earth magnetic eld nearly cancels in correspondence of the poles, while at medium and low latitudes the bulk of cosmic rays is cut out by the Earth magnetic eld, that for them acts like a mirror (see Fig. 3.7). The total rate of particle collected by the PAMELA telescope, averaged on the polar orbit and on the solar activity cycle, is 3.3 event/s, i.e. about event/day (mainly protons), ranging from 2.3 event/s to 4.2 event/s going from the period of maximum solar activity to its minimum. Averaging on the solar activity of the rst three years foreseen for the PAMELA ight ( ) we expect to collect in 10 8 s the numbers reported in Table 3.2 for particles, antiparticles and some nuclei. Not more than 10% of these events are expected to be complex, i.e. requiring the registration of more than 5 kb (40 kbit), the rest consisting of protons going through the whole telescope with no or poor interaction; the total information for them should not exceed 2.5 kb. The rate of fast triggers could be higher than the number of the particles going through the spectrometer; however the experience of the last balloon ights suggests 31

34 protons antiprotons electrons positrons He nuclei Be nuclei Cnuclei antinuclei limit (90% c.l.) 710,8 Table 3.2: Expected rates in PAMELA in 10 8 s that the number of fast triggers should be less than two times than that of the particles. From background simulations (see chapter 5) it can be concluded that also the instant trigger rate should not be overloaded neither by fake triggers nor by the crossing of the radiation belts at 60 {70 latitudes, the only possible exception being constituted by the crossing of the South Atlantic anomaly (see Fig. 5.5 and Fig We can indeed evaluate the information ow from PAMELA of not more than 2.5 kb/trigger, with no more than trigger/day, resulting in a total of not more than 10 Gbit/day of information. The way PAMELA will handle this information is reported in chapter 13. Figure 3.8: Satellite lifetime versus the F 10:7 solar activity index for circular orbits. Data based on assumed drag coecient of 2.1 and ballistic coecient of kg m,2 3.4 The spacecraft Resource{Arktica and the mission prole. The PAMELA experiment wiil be installed on the up{ward side of the Resource{Arktica satellite, that will be continuously oriented down{ward to the Earth during all its mission, in order to fulll a program of Earth surface observation. Furthermore the satellite will travel in a quasi-circular, about 700 km high, polar orbit. This is an optimal situation for the observation of cosmic rays: 32

35 the up{ward orientation of the PAMELA telescope on board of the satellite is the required direction for observing cosmic rays without interference with the Earth and keeping far away from the telescope acceptance the showers produced by quasi{horizontal cosmic rays on the terrestrial atmosphere; these showers are responsible of the strong increase of the background at low zenith angles; as above mentioned, the polar orbit maximizes the cosmic ray collection rate (important for antinuclei search), and also minimizes the geomagnetic cuto in a signicant portion of the satellite trajectory (important for all scientic issues related to the solar activity and to the terrestrial geomagnetic eects, see chapter 2); nally the high height of the orbit assures a long permanence of the satellite in orbit due to the very small eect of the atmospheric drag (see in Fig. 3.8 the permanence time in orbit for several satellite height and in dependence of a parameter connected to the solar activity); the stabilization of the satellite is obtained by magnetic devices to make the best use of this situation without depending from the possible shortness of fuel. The satellite is the upgraded version of th RESOURCE{01 satellites that the All Russian Science{Research Institute of Electromechanics (VNIIEM) of Moscow is regularly delivering in a solar syncronous polar orbit for a long duration program of Earth surface observation. A brief presentation of the VNIIEM and details of the satellites are given in the chapter 4 below. The mechanical and electrical interfaces of PAMELA with the Resource-Arktica satellite are described in the successive chapter 6. 33

36 34

37 Bibliography [1] M.P.De Pascale et al., J. Geophys Res., 98 (1993) 3501; R.L.Golden et al., Ap. J., 436 (1994) 769; P.Papini et al., Proc. 23rd ICRC (Calgary) 1 (1993) 579; G.Basini et al., Proc. 23rd ICRC (Calgary) 3 (1993) 773; C.Grimani et al., Proc. 23rd ICRC (Calgary) 4 (1993) 507 M.Circella et al., Proc. 23rd ICRC (Calgary) 4 (1993) 503. [2] R.L.Golden et al., WiZard-Related Balloon Program, Proposal to NASA NRA-OSSA-10 (1992) [3] R.Bellotti et al., 23rd ICRC (Calgary) 2 (1993) 635; R.Bellotti et al., Nucl. Instr. and Meth., A350 (1994) 556. [4] P.Carlson et al., Proc. 23rd ICRC (Calgary) 2 (1993) 504; P.Carlson et al., Nucl. Instr. and Meth., A350 (1994) 556 [5] Streitmatter, R.E., et al.: 1989 Adv. Space Res. 9, (12)65. [6] Beatty, J.J., et al.: Ap. J. 413, 268. [7] Mitchell, J.W., et al.: 1993a, Proc. 23rd ICRC, 1, 519. Mitchell, J.W., et al.: 1993b, Proc. 23rd ICRC, 2, 627. [8] Lawrence D.J. et al., Proc. 24rd ICRC (Rome), 3 (1995) 681; [9] Streitmatter, R.E., et al.: 1993 Proc. 23rd ICRC, 2,

38 36

39 Chapter 4 Spacecraft and Mission description 4.1 \Resurs{Arktika" SC purpose and tasks Resurs{Arktika is designed to receive and transmit prompt data on sea surface status, ice situation and meterological conditions in Earth polar regions, heliophysical data, information for Earth natural resources study, data on ecology and emergency, as well as to support digital data exchange between ground users. To provide safe, year{round navigation in Earth polar regions, in the North sea route primarily, the following parameters should be determined: sea surface status; meterological conditions in the specied regions; ice cover observation; ice elds size; ice age; ice cohesion; location of ice edges, fast shore ice frontiers, top and shelf glaciers; drift{ice and drift boundaries location; heavy sea roughness location; detection and observation of hurricanes, typhoons and other dangerous phenomena; determination of ecological situation. As the experience has shown, to determine the dynamics of status changing of hydro{meteorological and ice situation in polar regions, it is necessary to perform a 3{4 day cycle of continuous observation of the Arctic Ocean and a week{cycle of the Antarctic regions observation. It is possible to fulll the above mentioned task with promptness with the help of the specialized SC \Resurs{ Arktika" which is to be inserted into the sun{synchronous near{circle orbit with an average altitude of 690 Km and an inclination of 98.5 deg. This SC will make it possible with the help of additionally installed on{board instrumentation, to fulll the tasks having dierent aims in the interests of Russian and foreign customers. 4.2 On{board instruments composition The instrumentation installed on{board the \Resurs{Arktika" SC consists of informational (scientic) one and supporting (service) one The informational instrumentation \Severyanin" side{looking real aperture radar (RLS): the rst sensor designed to obtain information on sea surface status and ice situation; MSU{S2: multi{spectral scanner of moderate resolution with plane scanning designed for synchronous receiving imagery from RLS in visible spectrum and IR{band which will enable to increase the information on sea surface status and ice situation; 37

40 MSU{SK: multi{spectral scanner of moderate resolution with conical scanner having the same purpose as MSU{S2 in synchronous receiving imagery from RLS in visible spectrum and IR{band and being designed to obtain information on Earth natural resources; MSU{E: high{resolution multi{spectral scanner with electronic scanning designed for receiving high{ resolution imagery in visible spectrum for hydro{meteorology, Earth natural resources study and in emergency situation; Unied BISU{A: on{board information system designed to collect data own from on{board measuring instruments in a standard digit structure, data compression, memorization and transmission of ow over superhigh{frequency and ultra{high frequency{ band radiolinks; MZOAC UHF: radiometer designed to receive hydro{meterological data on ocean, atmosphere and ground surface; Data collection and transmission system (DCTS): designed to obtain hydro{meteorological parameters from sea, ice and ground automatic measuring platforms (AMP) and to transmit them to the ground; PAMELA Magnet Spectrometer (International collaboration): designed to register charged particles uxes and non{stationary phenomena investigation; Two{directional communication systems for ground users \ANTARES{R" (Germany) and \IRISAT" (Belgium) designed to support digital data exchange between users Main systems of the service instrumentation 3{axis attitude control system providing accuracy of SC attitude control in pitch and roll not more than 7 angular min and in yaw not more then 10 angular min. Angular rates max value in the mode of stabilization along each axis is not more then 0.005/sec; power supply system with voltage of 24{34 V and average power per 24 hours up to 800 W; data transmission system with carrier frequency of 8192 MHz and transmission rates of and Mb/s; thermal control system; command radiolink; TM system; antenna{feeder complex Structure \Resurs{Arktika" SC is presented in Fig. 4.1 and Fig As is shown from the gures, it consists of the following main parts: cylindrical pressurized body comprising 3 compartments where service systems are accommodated which provide SC functioning, and electronics boxes of information systems; external platform on which informational and service systems sensors, antenna, transmitters that tolerate operation in open space are located; solar array panels (SA). External instruments are connected by electric cables with instruments in the pressured{sealed connectors in a narrow belt of the body shell. Test process connectors to be used during SC electrical integrated test are located in the same area. The required temperature conditions of the instruments are provided by the thermal control system which comprises special coatings to be put on structural components and instruments cases, shield{vacuum insulation elements, a set of units to arrange gas ow inside the pressurized compartment, a fan with redundancy, a set of heaters and automatics box which controls the system instruments to the signals of temperature{sensitive elements located in dierent SC parts. External radiators and heat pipes are used on the external platform as well. \Resurs{Arktika" SC is inserted into orbit by Zenit (or Rus) launcher to which it is joined via the separation system by its rear part (i.e. by the part where PAMELA instrument is installed). 38

41 Figure 4.1: external view of the \Resurs{Arktika" SC. 4.3 PAMELA instrument design As mentioned in Section 4.2.3, PAMELA instrument is accommodated in SC pressurized and is installed on the instrumentation frame. Such xation makes it possible to perform SC electrical integrated tests when instruments cases are taken o, and hence, it makes it possible to replace separate instruments (and PAMELA as well) or their maintenance without a complicated operation of SC development. PAMELA frame design is the same as the main SC instrumentation frame and it seems to be its continuation. The joint between the frames is done in 8 points. As the common length of the two frames is rather big (4 m), on the level of their joint there are installed additional xation components limiting vibrations in the plane normal to the longitudinal axis of the cylindrical body. On PAMELA instrument faces plates are installed where the magnetic system and registering devices are xed. The electronics boxes are located on the instrument periphery and are xed to the frame load{bearing elements. On the low frame face (on the side of xation to the main frame) an intermediate panel is installed where there are electrical connectors of the cables connecting PAMELA instrument with SC devices and pressure{sealed connectors. For the purpose of cables length reducing, the number ofsuch panel may be2, 3 or 4. The instrument thermal conditions is provided by SC thermal control system. 4.4 Integration and tests SC integration to the plant is performed in the following sequence (see Fig. 4.3). 1. Integration of the body section (position 2), the external platform (pos. 1) and the instruments installed on it are performed on a special support. 2. The integrated SC fragment is moved and installed on the test-bed (pos. 7). 3. The frame with instruments (pos. 3) is installed on the body section (pos. 2). 4. PAMELA frame (pos. 4) is xed to the frame (pos. 3). On{board cables and ground cabling are mounted. PAMELA instrument should have a panel to which cables from SC instruments are joined. In such conguration SC undergoes integrated electrical tests. 39

42 Figure 4.2: Cross section of the \Resurs{Arktika" SC. 40

43 Figure 4.3: Scheme of the \Resurs{Arktika" SC for integration. 41

44 5. After the tests are completed the body section (pos. 5) is installed on SC, the frame is xed in the upper cross{section (Fig. 4.4) and the body section (pos. 6) is installed which is the PAMELA outer case. 6. SC structural integrity is checked. 7. Solar array panels are installed. 8. SC is taken from the integration test{bed and removed to the radio{technical test chamber. 9. SC is withdrawn to a special test{bed where SC mass and mass center coordinates are determined. 10. SC is set up in a horizontal position with the help of a manipulator and is prepared for transportation. 11. SC is packed in a transport container. 4.5 Transportation 1. The transport container is taken by a special car to an airport or a railway station. 2. SC is transported by air or by train to a launch site. 3. In case of transportation by air, SC is taken by car to a technical position where SC integrated checks are performed to conrm its preservation after transport, as well as structural integrity checks and pressurized compartment lling by nitrogen. 4. SC is again packed in a transport container and by a special railway carriage is taken to a launch position where it is integrated with the launcher. 5. The launcher together with SC is transported by a special assembly to the launch site and is set up in a vertical position. 6. Pre{launch checks are carried out and the launch is performed. Figure 4.4: Particular of the \Resurs{Arktika" SC cross section. 42

45 Chapter 5 Background conditions and expected exposition The orbit parameters of a spacecraft Meteor-3A are: 700 kms of height and 98 of inclination. The measurement conditions along a spacecraft orbit are dierent owing to the existence of the geomagnetic cut o. Therefore the measurement of antiparticle spectra at low energy is possible only at high latitudes, two times in each orbit during about 20% of a cycle time of the spacecraft. The study of particle trapped uxes is possible only in the Brazilian anomaly region on eight revolutions per day for about 10 minutes on each revolution. At high latitude the main background ux is due to galactic low energy protons and particles of the outer radiation belt. In this radiation belt the main source of background are low energy electrons. They increase the loading of all detectors and the false triggers due to coincidences with dierent particles. The very high background in the outer radiation belt could arrive to saturate and switch{o the detectors. In the inner radiation belt the main background consist of low energy protons up to 1 GeV. 5.1 Models used for background calculation For the estimation of the background conditions during the PAMELA orbit, the following basic data are used: Circular orbit of satellite: altitude 700 Km; inclination 98 Geomagnetic eld model with 6 harmonics Model of galactic cosmic ray (Russia, Standard) Model of albedo electrons (Russia, Standard) Model of electron and proton uxes in the inner and outer radiation belts (Russia, Standard; AP{8,AE{8 NASA). The instrument geometrical factors: { Total PAMELA geometrical factor: 23.5 cm 2 sr { Top detector S11/S12: 4059 cm 2 sr { Bottom scintillator detector S21/S22: 1118 cm 2 sr { TRD (1 layer): 2460 cm 2 sr { Top anticoincidence detector CAT: 2815 cm 2 sr { Lateral anticoincidence detector CAS: cm 2 sr { Silicon calorimeter: 6446 cm 2 sr Energy threshold: { Top detector: 20 MeV for protons and 3 MeV for electrons { Trigger: 80 MeV for protons and 50 MeV for electrons { Anticoincidences detectors and TRD: 20 MeV for protons and 3 MeV for electrons { Other detectors:80 MeV for protons and 50 MeV for electrons Temporal parameters of the instruments: 43

46 { Resolution time of coincidences: 10 ns { Dead time of the instrument: 10 ms { Prohibition signal duration of anticoincidence detectors: 50 ns Axis direction: vertical Movement of the satellite along orbit during one day is simulated as a step{by{step movement with a chosen constant temporal step. On each step geographical coordinates, Mac{Ilvain L and B coordinates, components of the geomagnetic induction vector, pitch angles and threshold rigidity are calculated. Using these values the integral intensities of electron and proton uxes are estimated. Then, under the known performances of Pamela spectrometer (geometrical factors, energy threshold and angular parameters), the expected counting rates of single detectors, counters and trigger system as well the eective expositions are determined. 5.2 Results of evaluations Exposition In Fig. 5.1 the temporal dependence of counting rate of the top detector along one satellite orbit during a day is shown. The detailed picture of background conditions in the top detector during a passage in the radiation belts (there are 8 such cycles in each day) is reported in Fig This gure shows that the zone of a quiet radiation conditions is about 60% of such cycles. The information on separate detectors allows to estimate the conditions of their reliable functioning. In Fig. 5.3 and 5.4 temporal dependencies of counting rates of the trigger are presented in cases of only real protons and electrons (Fig. 5.3) and of only random coincidences (Fig. 5.4). In Fig. 5.5 and Fig. 5.6 these dependencies are reported for one typical cycle, traversing through an inner radiation belt. The sum of these counting rates gives the complete counting rate for trigger signals. Reduction of the random coincidences is expected when in the trigger system is included the S13 scintillator too. A further reduction is possible by decreasing the resolution time of the coincidences block. In Fig. 5.7 and Fig. 5.8 temporary dependencies of the counting rates of both top and lateral anticoincidence detectors are shown. In Fig. 5.9 the temporal dependence of the background particle counting rate for the rst transition radiation layer is reported and in Fig it is reported the counting rate for the silicon calorimeter. It is clear that in some zones of the inner and outer radiation belts the counting rates can reach abnormal level for a reliable work of the detectors Numbers of registered particles and expected volume of the stored information For an estimation of expected number of antiprotons registered by PAMELA experiment, we consider the energy intervals shown in Table 5.1 with the correspondent p/p ratio deduced by the poor available measurements. Energy in GeV 0.2{ {1.0 1{10 10{30 >30 p=p 10,5 310,5 210,4 310,4 10,3 10,4 Table 5.1: Antiprotons/protons ratio in various energy ranges. The expected number of registered antiprotons in one year is calculated considering the galactic protons number in various energy ranges with a one{day exposition, then it is extrapolated for 1 year exposition and it is multiplied by the p/p ratio. The result is shown in table 5.2. For each energy ranges the reduction of registered particles due to a decrease of the real exposition is not more than 20%. This high eciency fulll the main goal of the experiment that is the antiproton study in primary cosmic rays. Similar results for total electron{positron ux of galactic origin are presented in table 5.3. Excluding from observations the zones of the inner radiation belt (for example by switching o the instruments, blocking the trigger circuit or passing a trigger signal through the scaled circuit), it is possible to estimate the total volume of the daily information. This average counting rate of trigger signals is about 3 Hz and considering an information volume of 2.5 kb for one event it is possible to receive about 650 MB of information per day. 44

47 Energy in GeV >30 Cosmic ray protons per day without the account of dead time Cosmic ray protons per day in view of dead time Cosmic ray antiprotons per year without the account of dead time Cosmic ray antiprotons per year in view of dead time Table 5.2: expected numbers of registered antiprotons and protons. Energy in GeV >30 Cosmic ray electrons per year without the account of dead time Cosmic ray electrons per year in view of dead time Table 5.3: expected number of registered cosmic ray electrons (e + +e, ) for 1 year of work. 5.3 Measurements in the radiation belts Protons with energy higher than several tens MeV are mainly contained in a particle ux of the inner radiation belt, trapped by the geomagnetic eld. The electron fraction is no more than several percents, 4 He nuclei are no more than several percents too and other nuclei are less. Therefore, the protons are the main source of background for the PAMELA instrument in the inner radiation belt. To estimate the opportunity of particle ux measurements in the inner radiation belt, areas of near{earth space with Mac-Ilvain coordinates L=1.12{2.0 and B<0.26 G are chosen. The background conditions are illustrated in Fig. 5.3, 5.4, 5.5 and 5.6. The number of particles (protons) trapped by geomagnetic eld registered during one day is reported in Fig Here, it is shown the number of registered protons in the case of no dead time, with a dead time of 10 ms, with the switching o of the lateral anticoincidences and with the switching o of the top anticoincidences. Without With dead With switched o With switched o dead time time 10 ms lateral AC lateral and top AC Table 5.4: number of protons in the inner radiation belt per day The calculated data allow to estimate the daily volume of informations, relating to the satellite trajectories, belonging to the inner radiation belt only. If the main trigger is used, the daily volume of informations is about 150 MB without taking into account the random coincidences and about 300 MB with the random coincidences (in the inner radiation belt the two contributions are comparable, as shown in from Fig. 5.3{5.6). 45

48 5.3.1 Expected measurement results in the radiation belt Nuclei In table 5.5 estimations of the energy threshold with the PAMELA instrument for various nuclei are presented. N. Nucleus E min, MeV E min, MeV/n R min,mv 1 p d t He He Li Be B C N O F Ne Al Ar Ca Ti Cr Fe Zn Table 5.5: energy thresholds for nucleous measurements in PAMELA. The results of numerous measurements in the inner radiation belt have shown the existence of signicant proton ux with energy about 1000 MeV on drift shells L=1.12{1.6. As the protons at such energies are stably trapped and accumulated by the geomagnetic eld, it is necessary to search the existence of trapped nuclei uxes in the inner radiation belt up to rigidities corresponding to protons at 1 GeV kinetic energy. This value is higher than the rigidity threshold for nuclei up to aluminium, therefore using PAMELA instrument it is possible to study the nuclear composition and the nucleous energy spectra in the inner radiation belt in the range 50{250 MeV/nucleon. At the present no data are available for nuclei at these high energies. Supposing a similar composition in the radiation belt and in the galactic cosmic rays, it is possible to estimate the numberofnuclei registered by PAMELA instrument. For example, it would be possible detect 10 2 {10 4 oxygen nuclei per day. Electrons Measurements of positron ux in the radiation belt, carried out by Russian experiments on satellite, have shown that, at high energy, the electron ux is about several percents of the proton ux. Therefore it is expected 10 3 {10 5 electrons per day depending on the PAMELA trigger in the radiation belt. Antiprotons The antiproton expected ux in the inner radiation belt can reach the value of 10,5 times the proton ux. It corresponds to 1{100 registered antiprotons per day, depending on the kind of trigger in the radiation belt. For example, using the second level trigger (see section 12.2), registering the total energy deposit in the calorimeter, we would obtain the real antiproton count per day as shown in Table 5.6. The background of antiproton events using the calorimeter second level trigger comes mainly from multiple events in the calorimeter itself. The number of background events is evaluated and shown in Table 5.7. Daily information volume, connected with registration of such imitations, can reach 250 MB. Therefore ground process selection of real antiprotons is required. 46

49 Trigger Basic 1 GeV 1.5 GeV 1.9 GeV Number of antiprotons per day Table 5.6: Antiproton count per day in the radiation belt using the basic trigger and the second level trigger based on a threshold in the calorimeter energy deposit. Type of energy deposition energy deposition energy deposition imitation 1 GeV 1.5 GeV 1.9 GeV two particles three particles Table 5.7: number of antiproton background events in the radiation belts in the case of second level trigger based on the total energy deposit in the calorimeter. 47

50 counting rate,s time,s Figure 5.1: counting rate on the orbit for scintillator detector S11. 48

51 10 7 outer radiation belt 10 6 outer radiation belt inner radiation belt outer radiation belt outer radiation belt counting rate,s high lattitude high lattitude 10 2 equator equator equator time,s Figure 5.2: Counting rate during one cycle of the orbit for scintillator detector S11. 49

52 counting rate,s time,s Figure 5.3: Counting rate on the orbit for trigger. 50

53 counting rate,s time,s Figure 5.4: Counting rate on the orbit for random coincidences. 51

54 inner radiation belt counting rate,s high lattitude high lattitude 10 0 equator equator equator time,s Figure 5.5: Counting rate during one cycle of the orbit for trigger. 52

55 counting rate,s outer radiation belt outer radiation belt high lattitude inner radiation belt outer radiation belt high lattitude outer radiation belt equator equator equator time,s Figure 5.6: Counting rate during one cycle of the orbit for random coincidences. 53

56 counting rate,s time,s Figure 5.7: Counting rate on the orbit for top anticoincidence counter. 54

57 counting rate,s time,s Figure 5.8: Counting rate on the orbit for lateral anticoincidence counter. 55

58 counting rate,s time,s Figure 5.9: Counting rate on the orbit for one layer of TRD. 56

59 counting rate,s time,s Figure 5.10: Counting rate on the orbit for calorimeter. 57

60 58

61 Chapter 6 Interfaces The system of electrical and electronic interfaces of the PAMELA telescope executes the following functions: connection of instrument to the system of the primary on{board power supply; control of instrument modes of operation by telecommands (TLC) from the Earth and from on{board program{timing device (BPTD); reception of time code from the system of on{board standard time and frequency (SBTF) for timing of scientic information; reception of scientic information from detectors, possible compression, timing, formatting, accumulation and transferring on{board information control system (BICS) for transmission to the Earth; reception of information about parameters of instrument condition and inclusion of this information into a ow of scientic information; interface of instrument with slow telemetry system. 6.1 Power Supplies for the PAMELA devices In this section we will describe the interface between the power supply of the russian satellite and the PAMELA power supply system. Meanwhile, we will give also a detailed description of the proposed scheme for the PAMELA power supply system Introduction Dierent PAMELA subsystems have dierent power supply requirements. In detail PAMELA needs: High Voltage Power Supplies (HVPSs) with limited current for the Photo multipliers (PM) of the TOF and of the anticoincidence system, and for the Transition Radiation Detector (TRD) (respectively V for the PM and V for the TRD). Power Supplies (PS) to deplete the silicon sensors ( V). Low voltage Power Supplies (typically +/-5 & +/-12 V) for all the sub-detectors systems and for the general processing and control electronics (pre-processing DSP boards, main CPU, telemetry system...). Low Voltage, low noise PS for the preamplier readout electronics of the detectors (i.e. 2 V oating PS for the front{end electronics of the silicon detectors). independent backup power supply for the slow control CPU. The above PSs are derived from the main PS system of the Russian satellite. In what follows each PAMELA PS will be deal in detail Main PS system and interface Each PS of PAMELA is derived from the 27 V - Main power supply of the Russian satellite (MPS), according to the scheme in g The fuse protected and EMI/EMC ltered 27 V power supply is issued through a relay board to the dierent PAMELA PS sections, where it is again EMI/EMC ltered. The connection cables from the relay board to each PS section is fuse protected at both ends. The power supply are actively controlled by the slow control CPU. The total available power is approximately 300 W. 59

62 Figure 6.1: Schematic description of the interface between the main power supply of the russian satellite and the power supplies of the dierent PAMELA devices HV Power Supplies for PM and TRD The main design guidelines for the HVPSs are: minimize the out-of-run probability of each PM and TRD due to electronics failures; provide individual tuning capability of the voltage for each sensor. The tuning range is selected to optimize the sensor gain ( V for the PM and V for the TRD). The same solution is proposed for both the PM and the TRD, and it is shown in g. 6.2 with reference to the PM. It is based on the use of two HV generators (1 + 1 spare) for up to 10 PM (TRD), each one provided with a linear voltage regulator. For the HV generator two solutions are considered (in any case a spatially-certied Figure 6.2: The solution adopted to generate and distribute the High Voltage power supplies for the photomultipliers and the TRD. commercial PS will be used) : a y-back PS with a push-pull zero-voltage-switch driving section and a toroidal transformer. The drawback of this solution is the eciency that, for a maximum current of 500 A (each PM sinks up to 20 A) can't be greater than %; 60

63 a voltage multiplier based on voltage{doubler diode cells. This solution has good performance in terms of eciency, but requires a high degree of redundance to guarantee an acceptable reliability. The regulator section (g. 6.3) is based on a MOS transistor with a parallel resistor to limit the drain-source voltage (Vds) to 200 V. The MOS gate-source voltage (Vgs) is obtained by rectifying a square wave signal with a voltage-doubler circuit. The amplitude of the square wave is controlled according to a feedback signal from the sensor. The redundance level of the scheme in g. 6.3 was selected among dierent redundance solutions to Figure 6.3: Blockscheme of the regulator section used for the high voltage power supplies of the photomultipliers and the TRD, with the redundances adopted to minimize the out{of{run probability. minimize the out{of{run probability of each sensor. In detail, the linear regulator is not critic for the reliability because a failure in this section only prevent from the gain regulation of the sensor but doesn't exclude the sensor itself. In fact the sensor, thanks to the resistor in parallel to the MOS transistor, will work in this case with a gain corresponding to the lower level of the tuning range. The HV generator is the critical block of the chain. The conguration of a duplicated HVPS for 10 PM (TRD) was selected as the best trade-o between the redundancy level and the single PM (TRD) out-of- run probability. The critic components such as disjunction diodes and HV capacitors are redounded in order to account both for an open and for a short circuit failure PS for the Silicon detectors Each one of the 18 PAMELA tracker ladders (described in chapter 10) requires a reverse bias in the range V (with respect to the PAMELA ground), plus eventually a bias voltage in the range 0-5 V to control the FOXFET gate on the junction detector side. The arrangement of the PS for each ladder is shown in g. 6.4, where each HV generator is designed following the same solutions and criteria described in the previous section. In particular the duplicated arrangement of the PSs for each ladder was chosen as a good trade-o between redundancy level and single channel out-of-run probability due to electronic failures. Disjunction diodes and HV capacitors are again redounded in order to account both for an open and for a short circuit failure. Power output and power dissipation are negligible Low voltage PS for the tracker, TRD and calorimeter readout electronics General remarks The preamplier circuits of the Tracker, TRD and Calorimeter Detectors are characterized by the high required Signal to Noise ratio and by the high sensitivity, in such away that the power supply of the circuit must be free from switching noise and well stabilized. The more sensitive part is the VA1 chip of the Tracker readout. The VA1 uses as input stage a non dierential scheme, so that the long term stability must be better than 1% to prevent changes in the gain of the circuit, and the switching noise less than 1 mv PP. The power supply scheme adopted for the tracker will be used also for the TRD and the Calorimeter, with dierent value of the voltage, depending on the selected preamplier chip. In the following paragraph we will describe the design 61

64 Figure 6.4: Block scheme of the power supplies used to deplete the silicon sensors. of the tracker's low voltage power supplies. The TOF and anticoincidence board do not need low noise power supply due to the intrinsic gain of the PM which directly insure a big signal (' 50 mv). System description As described in chapter 10, the PAMELA tracker is arranged in six planes of silicon sensors. Each plane is organized in three ladders made up of two double-side silicon sensors. Each ladder is subdivided in two sides (junction and ohmic side) which have their own readout electronics on dierent sides of the same hybrid circuit and on a dedicated ADC card. The hybrid and the ADC card for each semi-ladder require a 2 V with respect to a reference voltage, which is the HV level of the silicon detector PS for the ohmic side semi-ladder, and the PAMELA ground (directly or via 2 antiparallel diodes) for the junction side. The required long term stability of the PS is 50 mv. The 2 V PS must be carried out with a bipolar technology because the power consumption of the VA1 chips is not symmetric. In detail in the selected bias conditions (section ), each VA1 chip has a power consumption of 0.7 mw/ch, that is 0.7 W/side of a ladder (1024 channels). The total current drawn by the VA1 chips for each side is about 350 ma in total, 230 ma from the -2 V PS and 115 ma from the +2 V PS. For the tracker, the modularity of one PS for each plane was selected for eciency reasons, with a redundance of two for reliability purposes. In the next section the arrangement of the PS for one plane will be described and discussed in detail. Detailed description of a three-ladder readout-electronics PS For each plane the PS is arranged as shown in the block scheme of g The transformer is carried out using a toroidal core with one centre-tapped primary winding and four centre-tapped secondary windings (one winding for each one of the power groups of the ohmic-side readout electronics of the three ladders, and one common winding for the remaining power groups of the junction-side). The primary winding is driven by a push-pull resonant (around 75 khz) MOSFET circuit. Each one of the secondary windings has a full bridge rectiers, a compensating inductors (since the feedback is derived only from the power group which provides power for the junction side), a LC lter carried out with pot-core inductors (to lter switching diode spikes), a common mode lter made with a toroidal multi winding inductor, and a low-drop regulator. The low-drop regulator is a custom chip which allows an automatic, passive load-control. The essential circuit for one power group is reported in g In normal operating conditions the operational amplier A drives the regulating transistor Q1 so that the voltages on the inverting and the non-inverting pins of A are equal, while the operational amplier B has no feedback circuit because diode D1 is reverse-biased. When over current on resistor R1 occurs (normal current values are around 200 ma, the threshold current to switch the fold-back protection is 400 ma), B and D1 brings A in positive saturation limiting the current on Q1 to 10 ma (which is a not dangerous value also for a short-circuit load). The advantage of this circuit with respect to a traditional ON-OFF fold-back protection is that it automatically senses if the short-circuit is over and automatically restores the normal-run conditions, 62

65 Figure 6.5: The arrangement of the low voltage power supplies used for each plane of the tracker system. Figure 6.6: Electrical scheme of the low drop linear regulator for the low voltage power supplies of the tracker. 63

66 while the ON-OFF protection requires an active control of the load-status, that is a drawback with respect to the circuit reliability. The PS schematically reported in g. 6.5 is duplicated for reliability purposes up to the dashed line, which is the point where the spare units are connected to the primary ones. As mentioned in the previous sections, the modularity of one PS for each plane of silicon sensors was selected for eciency reasons. The PS described in this section can achieve a 60-65% eciency for an output power of about 6 W, which is the power consumption of the readout electronics for one plane of silicon sensors (for each side of a ladder: 0.7 W for the VA1 chips, and about 0.3 W for the ADC, the control and the interface logic) V PS for the sub-detectors and the general processing and control electronics The processing and control electronics (pre-processing DSP boards, main CPU, telemetry system,...) and some parts of the sub-detectors' electronics will run with a standard 5 V PS. The 5 V PS will be obtained from the 27 V MPS system using a VICOR module MI J20MZ, which is a 28 V input - 5 V output, 25 W, 80-90% eciency, MIL-SDT-810 certied DC/DC converter. Three MI J20MZ modules will be used (1 + 2 spare) for reliability purposes. Analogue module are available for -5 V and +/-12V (if needed). 6.2 Control of operation mode of PAMELA devices A command{measuring system (CMS) COMPARUS is used in \Resurs{Arktika" S/C. On{board commutation unit (BCU) accepts codes of telecommands (TC) from CMS and sends on{board control commands with the following parameters: polarity: positive; duration: 0.1{1.0 s; amplitude: 24{34 V; load current: not more than 1 A; The common commands bus is connected to \{" of on{board power supply. The total number of commands is 125. they eectuate the operation control for each device: TOF system: 25; TRD system: 25; Tracker: 25; Calorimeter: 30; Trigger logic: 10; Information processing system: 10. TC control of PAMELA elements turn on/o the various systems, change thresholds, switch on Test & Calibration modes etc. BPTD is used to activate time programs of PAMELA work in the orbit. BPTD forms sequences of commands starting from some initial moment of a time. The time of action will be dened by scientists some days before execution. 6.3 Reception of time code The \Resurs{Arktika" structure includes a system, to synchronize all the actions executed on the S/C in ight with a time. SBTF produces time signals in a kind of pulses of various frequencies with high accuracy and stability (10,8 s) and in a kind of serial 17{bit binary code of time, containing values of hours, minutes and seconds. A separate pulse, marking a day (24 hours) is produced as well. SBTF provides an opportunity of time correcting with ground stations and synchronization with UT. The code of time is transmitted by three signals with separate outputs: code \0" (K0); code \1" (K1); 64

67 end of a code transfer (KK). The signal frequency is 50 Hz starting from the highest bit (hours). The beginning of time code transfer coincides with pulse, designating \second label" (frequency of pulses 1 Hz) (see Fig. 6.7). Figure 6.7: The timing of time code. The millisecond mark is available for instrument. All SBTF signals have a transformer output and are capable to work on a load of 100. The pulses have positive polarity and the following parameters: amplitude: 6{9 V; duration: 3{6 ms. 6.4 Central processing and Slow Control Unit The central computing system is the control and acquisition unit for data acquisition, processing, storage and transfer of information (to be transmitted to the earth) to BICS. The central computing system is also the main controller for all o{line and slow control operations of the scientic payload included the correction of on board scientic and service software. The central CPU (CCPU) is the top level of the control and data acquisition network and the main interface to the Russian Telemetry. It contains also the mass buer memory for the data and the mass memory for all programmable devices in the detector hardware. Central station fullls the task of general control in weak synchronized mode, with CPU redundancy of the peripheral control and acquisition stations. Each peripheral station is dedicated to one sub-detector, and is duplicated for redundancy. The link between the sub-detectors electronics and the peripheral station is a redundant serial link in a star architecture so that the sub-detector read out electronics is not aected by single point failure. If one of the peripheral station fails to operate, the CCPU switches the channel on the duplicate link. The data handling is based on a master-slave conguration with time out procedure. Lower level electronics (front-end units to peripheral station) starts its data processing tasks in presence of a First Level Trigger and wait for a data request. After a maximum time interval the sub-detector controller cancels all the acquired data and starts again the acquisition phase. Second Level Trigger Units works in parallel with the main data acquisition link and have their own peripheral station units so that any failure on the parallel processing units does not aect the main data link. The CCPU system is the most critical device in Pamela Payload so that it will be fully qualied and produced by a spatial qualied industry. The main characteristics are: a redundant CPU and internal network; a secondary backup power supply; mass Memory greater than 2GB; an internal transfer rate of 18/31 Mb/s; redundant EPROM FLASH memory to allow in orbit reconguration of the software. 65

68 The use of a command measuring system channel (CMS) during control session in the Russian control room is foreseen in order to transfer codes for the software correction in ight to the CCPU. The connection between the CCPU and the satellite system will be carried out on a special developed I/O card that will be controlled by a program. The mass and power characteristic are presented in the general mass and power budget as maximum allowed quota. Based on preliminary estimations the daily volume of scientic information from the PAMELA telescope can reach 1GB. In order to transfer to earth such avolume of information in one or two daily sessions, it is necessary to use a fast telemetry channel of BICS (15.36 Mb/s). A complex buer subsystem unit is necessary in the communication line of the computing system BICS to provide a transfer with such a speed. An alternate solution can be a data compression unit which can allow the use of a channel of 1.28 Mb/s (Fig. 6.8). Figure 6.8: information transfer to the ground. 6.5 Telemetry system for the slow control of the apparatus A command{measuring system (CMS) COMPARUS includes the system of measurements (slow telemetry system). It accepts from instrument the analogous parameters: temperatures of dierent devices, voltages and currents of circuits, dry contacts are indicating the state of instrument and its parts. Analogous signals to be measured: voltage: 0{6.3 V; polarity: positive; accuracy: 3%; number of signals: 20; number of dry contacts:

69 Part II Detectors in PAMELA 67

70

71 Chapter 7 The scintillation counter hodoscopes In the PAMELA telescope are included several scintillation counter hodoscopes. Five of them (S11, S12, S13, S21, S22) are used for the measurements of the Time of Flight (TOF) of the going through particle, and a sixth one (S3) at the bottom of th telescope, under the imaging calorimeter, is used for a fast information on the penetration of the particle. Furthermore two set of scintillators (the Top Anticoincidence Counters { CAT { and the Side Anticoincidence Counters { CAS {) give a fast information on the contamination from particles entering the spectrometer volume from directions outside the geometrical acceptance of the telescope. 7.1 Requirements and design goals for the time{of{ight scintillator system The PAMELA TOF system will be made up of ve layers of scintillation detectors, two at the top of the instrument (S11, S12), one just above the magnet (S13), and two just below the magnet (S21, S22). A sixth scintillator layer (S3) will be located below the calorimeter, primarily as a penetration detector. The dimensions of the scintillators correspond to the geometrical limits of particle tracks through the spectrometer, projected to the locations of the scintillator planes. The scintillator system must meet the following major goals: 1. Detect incident particles passing through the aperture of the spectrometer so that a fast event trigger can be generated. This requires high detection eciency (essentially 100%) for singly charged, minimum ionizing particles and a rapid time response so that a coincidence time of at most a few tens of nanoseconds can be used. 2. Reject albedo particles at a level of better than one in 10 8 by identifying \upward" moving particles. This requires sucient timing resolution to separate downward moving and upward moving particles travelling at nearly the velocity of light by just over 5.5 standard deviations. 3. Dierentiate protons from positrons and electrons from antiprotons at a level of at least one part in 10 4 from the lowest energies measurable by PAMELA up to the point where the calorimeter and TRD become eective. While this depends on the nal performance of the calorimeter and TRD, we must assume that the TOF system should provide a separation of about 4 standard deviations between protons (antiprotons) and positrons (electrons) up to a momentum of at least 1.0 GeV/c. 4. Identify particles which penetrate the calorimeter, and measure the \leakage" of shower particles out of the bottom of the calorimeter. This requires that the penetration detector, S3, have a high detection eciency for singly charged, minimum ionizing particles and rapid enough time response to provide a coincidence signal in a time window of a few tens of nanoseconds. The requirements of high detection eciency and rapid time response are easily met by detectors using conventional plastic scintillators, 1 cm thick, coupled to fast photomultiplier tubes (PMTs). Similarly, the electronics required to meet these goals are conventional although care is required for a low{power implementation. The rejection of albedo particles is also straightforward. The distance between the upper and lower time{ of{ight layers is about 88 cm. The dierence in ight times for a downward moving and an upward moving particle at the speed of light is 5.9 nanoseconds. Thus, albedo particles can be rejected eciently by a TOF system with a resolution (sigma) of 1 nanosecond. This is easily achieved. The requirement that the TOF system be able to distinguish protons (antiprotons) and positrons (electrons) is somewhat more stringent. Over an 88 cm ight path, a TOF system with a timing resolution (sigma) of 69

72 200 picoseconds will give a four standard deviation separation between protons and electrons at 1.1 GeV/c. In the past few years, TOF systems with this level of performance have become relatively common and we are very condent that a resolution of at least 200 ps will be achieved. In fact, the minimum design goal of the PAMELA time{of{ight system is 150 ps, which will extend the useful range of the TOF to over 1.3 GeV/c. As discussed below, we expect the actual performance of the TOF to considerably exceed this design goal. (It should be noted that some care should be used in interpreting results quoted in the literature since the timing resolution quoted is often that of a single detector rather than that of a full TOF system. The single detector measurement will be a factor of 1.41 better than the ight time resolution of a system made up of two of the same detectors.) In order to assure that the performance goals of the PAMELA time{of{ight system are met, all aspects of the design must be carefully considered. We have developed an initial baseline design for the TOF based on our experience and our understanding of the state{of{the{art in both scintillators and photomultipliers. This initial design is presented below. There are some interesting alternatives to this design which may oer some advantages if sucient performance can be realized. The nal detailed design of the TOF will be arrived at after an extensive series of simulations and tests both of the baseline design and of these alternatives. 7.2 Experience in TOF design and fabrication Members of the NASA/Goddard Space Flight Center and University of Siegen research groups have been involved in the design, fabrication, and analysis of time{of{ight systems since This work includes the development of high performance TOF systems for balloon{borne cosmic{ray experiments including LEAP [8], SMILI [3], MASS, IMAX [5, 6], TS93, CAPRICE, TIGER, and ISOMAX [9] and for accelerator based experiments such as Lawrence Berkeley National Laboratory experiments E859H and E938H [2] and Brookhaven National Laboratory experiments E878 and E896. The design of the PAMELA time{of{ight system draws on this experience base. In particular, the photomultiplier tubes (PMTs) and scintillator selected for PAMELA are the same as have been used in IMAX and ISOMAX. The PMT bases and light pipes, while not identical to those used in IMAX and ISOMAX, are similar in concept. The timing and logic electronics are based on those under development for the ISOMAX long{duration balloon ights. IMAX utilized two TOF layers, each made up of three cm 3 panels of Bicron BC{420 plastic scintillator viewed through twisted{segment adiabatic light{pipes by Hamamatsu R2083 photomultiplier tubes. In the IMAX balloon ight, this system provided a ight time resolution of 122 ps for relativistic singly{charged particles [7]. As discussed below, it is expected that the PAMELA TOF system will exceed this performance. 7.3 The time{of{ight system General description As noted above, the PAMELA TOF will be made up of ve layers of 1 cm thick fast plastic scintillator arranged in three groups. Since the TOF system will be used to provide the basic experiment trigger, the dimensions of the scintillator layers are dictated by the projected trajectories of charged particles passing through the spectrometer. In order to maintain the full geometrical factor of the instrument (25 cm 2 sr) the top layers are 45 cm wide in the bending view and 40 cm wide in the non{bending view. The middle and bottom layers are 17 cm wide in the bending view and 15 cm wide in the non{bending view. The penetration detector is 36 cm wide in the bending view and 32 cm wide in the non{bending view. Each TOF layer will be made up of rectangular strips of scintillator with the number and dimensions of the strips chosen to give ecient coupling to the photomultiplier tubes. Each strip will be viewed at both ends by photomultiplier tubes coupled through adiabatic light guides. The timing layers will all be made up of Bicron BC{420 scintillator (rise time 0.5 ns, decay time 1.5 ns). S3 will be made of Bicron BC{404. The S11 and S12 layers will each be composed of four scintillator strips, each 45 cm long, 10 cm wide and 1 cm thick. The long axis of the strips will be in the bending plane of the spectrometer. This eliminates any interference between the TOF system light{guides or PMTs and the TRD electronics. In the non{bending view, the S11 layer will be slightly oset (1.5 mm) to one side and the S12 layer will have the same oset in the opposite direction. The resulting relative oset of 3 mm will insure that there is no loss of geometry factor due to gaps between the scintillator strips. The S13 layer, just above the magnet will be made up of a single 17151cm 3 scintillator. Light{guides will be coupled to each of the 17 cm ends of the scintillator. This eliminates any interference with the anticoincidence detectors surrounding the magnet. 70

73 The S21 and S22 layers will each be made up of two scintillator strips, each cm 3. The long axis of the strips will be in the bending plane to eliminate interference with the anticoincidence detectors. As with the top scintillators, the S21 and S22 layers will have a relative oset of 3 mm to eliminate the eect of gaps between strips. Using two TOF layers in the outer groups insures that the TOF system is single{fault tolerant without signicant loss of timing resolution. Also, two completely independent TOF measurements are made between these groups (e.g. S21{S11 and S22{S12), giving a factor of 1.41 improvement in the instrument timing resolution over that of a single pair of layers (strips). In normal operation, both layers in each outer group will be used in the trigger, together with S13. Since the layers in each outer group are slightly oset with respect to one another, there will be no dead area in the experiment geometry due to gaps between TOF strips. The penetration detector, S3, is made up of a single cm 3 scintillator. This scintillator is viewed by eight small PMTs coupled through relatively simple \sh{tail" light guides. S3 is not used in the experiment trigger, but serves to tag penetrating particles or showers Photomultipliers The choice of appropriate photomultiplier tubes involves considerations of both photocathode area and performance characteristics. In order to achieve the greatest eciency in collecting scintillation light, the eective photocathode area of the PMT should equal or exceed the exit area of the scintillator. In PAMELA, the exit areas of the TOF detectors are 10 cm 2 for S11 and S12, 17 cm 2 for S13, and 8.5 cm 2 for S21 and S22. This means that for typical PMTs with round photocathodes, the ideal diameters would be 3.6 cm (1.4 in) for S11/S12, 4.6 cm (1.8 in) for S13, and 3.3 cm (1.3 in) for S21/S22. Thus, a nominal 3.8 cm (1.5 in) PMT with a typical photocathode diameter of 3.4 cm (1.34 in) would be a good choice for the S11/S12 (with less than 10% loss of light collection) and S21/S22 scintillators. A nominal 5.1 cm (2 in) diameter PMT with a typical photocathode diameter of 4.6 cm (1.8 in) would be an ideal choice for S13. For use in the TOF system, photomultiplier tubes must exhibit excellent pulse characteristics including fast rise and fall times, low electron transit time spread, and little dependence of the transit time on the position of an incident photon on the photocathode. In addition, the PMTs must have reasonable quantum eciency, relatively low dark noise, and good amplitude (energy) resolution. Finally, the mechanical structure of the PMTs must be able to withstand launch loads. Most high performance TOF systems built to date have used PMTs with a time{compensated design for collecting the photo{electrons at the rst dynode and a linear{focused dynode structure. The best current PMTs using this construction have rise times on the order of 1 ns and transit time distributions with standard deviations on the order of 300 ps. Unfortunately, these high performance PMTs are not available in a full range of diameters. Currently, there are excellent linear{focused fast{timing PMTs available in 5.1 cm and 2.5 cm envelope diameters (4.6 cm and 2.0 cm photocathode diameters) but not in a 3.8 cm diameter. (There are also high performance timing PMTs with envelope diameters less than 2.5 cm.) As a result, if linear{focused PMTs are used, the best match to the PAMELA TOF scintillators is obtained if 5.1 cm PMTs are used throughout. In the past several years, there has been a great deal of interest in the performance of the Hamamatsu \ne{ mesh" PMTs, largely due to their insensitivity to magnetic elds. The dynode structure of these PMTs is made up of a parallel stack of plates made of a ne metal mesh. These plates are parallel to the (at) photocathode and span the full photocathode diameter. The anode has the same dimensions as the dynodes. The resulting tubes are very short compared to conventional PMTs with similar gain. The electron ight paths are nearly straight from the photocathode through the dynode stack and the transit time exhibits little dependence on the position of the incident photon(s). Although the rise time of these PMTs is about a factor of two greater than the best timing PMTs, good results have been obtained using these PMTs in TOF systems. The TOF systems developed by GSFC for the LEAP, MASS, TS93, and CAPRICE used Hamamatsu R2490 (5.1 cm) PMTs to give timing resolutions of the order of 200 ps in ight. Fine mesh PMTs were also used in the TOF developed for the 1995 BESS cosmic ray experiment [12] and gave a resolution of about 100 ps with 2 cm thick, 10 cm wide scintillators. In addition, beam tests of an accelerator TOF made up of narrow scintillator bars cm 2 in cross section viewed by R3432 (2.5 cm) ne{mesh PMTs gave resolutions of better than 100 ps [1]. Despite these results, the ne{mesh PMTs have some serious drawbacks for use in PAMELA. Principal among these is the relatively long rise time of about 2.5 ns (for a 5.1 cm PMT). This means that more correction must be applied to the data to compensate for amplitude dependent time slewing. Also, compared to a conventional PMT with the same envelope diameter, the eective photocathode diameter of the ne{mesh PMTs is smaller. The photocathode diameter of the new R5924 (5.1 cm) ne{mesh PMT is only 3.9 cm (and the photocathode diameter of the older R2490 was only 3.6 cm). Although a 3.8 cm ne mesh (R5946) PMT is available, its photocathode is only 2.7 cm in diameter. Thus, 5.1 cm PMTs would still be required for all of the PAMELA TOF scintillators but use of the ne{mesh would result in a 30% loss of collected light in S13. In addition, the quantum eciency of the ne{mesh PMTs is less than 80% of the best linear{focused 71

74 timing tubes. Finally, there has been some concern about the ability of the R5924 (5.1 cm) ne{mesh PMTs to withstand launch shock and vibration loads, although Hamamatsu has tested the R5946 (3.8 cm) and R5505 (2.5 cm) PMTs to shocks of 75 g for 11 ms and vibration of 5 g at 1.5 mm p{p and 10{500 Hz. Another very interesting PMT from Hamamatsu that has been considered for PAMELA is the R5900, part of the the newly developed \metal channel dynode" series. These are very compact PMTs which were originally developed (as the R5600 with a 0.8 cm diameter cathode) to replace solid{state photodiodes in some applications. The metal channel dynode tubes are less sensitive to magnetic elds than conventional tubes, and can tolerate elds of a few mt without shielding. The R5900 is the largest tube of this type developed to date. It oers a cm 2 (3.2 cm 2 ) photocathode area in a cm 2 square cross{section envelope. The R5900 has a rise time of 1.4 ns and a transit time spread of 281 ps (sigma), suggesting that it may be a very useful tube in timing applications. Researchers in Italy and the United States are currently evaluating the performance of the R5900 in the TOF system for AMS (the Alpha Magnet Spectrometer). While the ruggedness of the R5900 has not been established, Hamamatsu has tested the R5600 to the same shock and vibration levels quoted above for the ne{mesh PMTs. The primary advantages of the R5900 compared to conventional PMTs of comparable performance are that it is very compact and relatively insensitive to magnetic elds. It also operates at a nominal high{voltage of 900 V compared to 2{3 kv for conventional PMTs. The principal drawback is that at least three times the number of R5900s would be required to equal the performance of 5.1 cm diameter linear{focused PMTs. Although it would be possible to congure the TOF with narrower scintillator strips, this would be mechanically complex and would require over three times as many electronics channels unless a matrixing scheme were employed in which signals from dierent (non{adjacent) strips were sent to a single electronics channel. Thus, a more practical solution to an R5900 based TOF would be to view each strip with multiple (3 for S11/S12 and S21/S22) PMTs at each end. The PMT signals could be sent to individual electronics channels (still requiring three times the number unless matrixed) or the signals from the PMTs on each scintillator end could be \ganged" through a passive or active network and supplied to a single electronics channel. If the signals were ganged, then the same number of electronics channels would be required as for conventional PMTs. However, unless the PMTs are well matched in transit time and gain and unless pulse reections are suppressed (possibly by back terminating the connecting cables) the performance of such a system will be degraded. An additional drawback of the R5900 is that its quantum eciency is about 80% of the conventional 5.1 cm PMTs under consideration. Finally, the R5900 is a new, relatively unknown, device while the conventional PMTs are fully developed and their characteristics are well understood. There have been a number of reports that the performance of the metal channel dynode PMTs is quite variable from unit to unit and that only the good ones really have the required performance. This suggests that the metal channel dynode tubes are not yet a mature technology. Considering all of these factors, 5.1 cm diameter linear{focused photomultipliers have been chosen as the baseline for the PAMELA time{of{ight system. We have considered a number of 5.1 cm PMTs including the Philips XP2020, XP2020/UR, and XP2282B, and the Hamamatsu R1828{01 (the Hamamatsu equivalent to the XP2020) and R2083. Both Hamamatsu and Philips have been supplied with the nominal shock and vibration specications for the Resource Satellite launch. Hamamatsu believes that the R2083 and possibly the R1828{01 should be able to withstand these loads without modication. Philips is considering the question but thinks that it is possible that the PMTs under consideration may be able to withstand the loads as well. We anticipate using compliant mounting of the PMTs to further reduce the loads communicated to the PMTs. The XP2020, XP2020/UR, and R1828{01 are 12 stage high performance tubes with rise times of about 1.5 ns. They have been used very successfully in many TOF applications including the PBAR, SMILI, and TIGER balloon{borne cosmic ray instruments. The main drawback of these PMTs is that they have a relatively long (19 cm to the end of the plastic base) envelope. Their gain (210 8 at 3 kv) is also much higher than required for PAMELA. The XP2282B is an 8 stage tube, also with a rise time of about 1.5 ns. It has a shorter envelope (165 cm to the end of the plastic base) and a gain which is more in line with PAMELA requirements ( at 3 kv). The R2083 is an 8 stage PMT as well. The R2083 is one of the highest performance timing tubes available, with a rise time of 0.7 ns and a transit time spread of 0.37 ns FWHM. It has a short envelope (13.4 cm to the bottom of the glass bulb) and a gain which iswell matched to PAMELA requirements ( at 3 kv). R2083s were used very successfully in the IMAX balloon{borne cosmic ray instrument and will be used in ISOMAX. In addition, versions of the R2083 have been used in a number of TOF systems for use in accelerator experiments (for example see [10]) and it is considered to be one of the best fast{timing PMTs in existence. Based on the experience of the GSFC and University of Siegen groups and on the considerations discussed above, the Hamamatsu R2083 has been chosen as the baseline PMT for the S11/S12, S13, and S21/S22 detectors. The Philips XP2282B will be evaluated as an alternative. The Hamamatsu R5900 has been chosen as the baseline for the S3 penetration detector where its limited photocathode area is less of a drawback. As part of the development of the PAMELA TOF, we will continue to evaluate other alternatives and plan to test TOF 72

75 scintillators using the R Light guides The PMTs will be coupled to the scintillator strips using adiabatic light guides (or light pipes). considered two dierent types of light guide in developing the baseline design: We have 1. \sh tail" light guides, machined from a single piece of material, and 2. multiple segment light guides made up of several individual components. The sh{tail light guides are relatively compact and somewhat simpler to manufacture. Light guides of this type have been used in the TOF system developed for the 1995 BESS balloon{borne cosmic ray experiment, discussed above. However, the multiple segment type light guides have generally been shown to have higher potential performance. Based on the experience of the GSFC and University of Siegen groups, the multiple segment type has been chosen for the baseline PAMELA TOF. Each of the light pipes will be made up of several ultra{violet transmitting (UVT) acrylic \ber" segments. Ten segments with 1 cm square cross sections will be used in the S11 and S12 light guides. Eight 1 cm square segments, will be used in S21 and S22. The S11/S12 and S21/S22 segments will be individually formed (heated and bent) to make up a composite light pipe which will map the exit area of the scintillator onto a pattern whose largest dimension is less than the diameter of the active area of the photomultiplier (4.6 cm). The light guides will be constructed so that there is no interference between the light guides or PMT assemblies on adjacent layers. The light pipes will be coupled to the PMTs by a 5 cm long, 4.4 cm diameter UVT acrylic cylinder which will serve as an \expansion" volume to insure that the light from each of the segments has approximately an equal probability of illuminating the active area of the PMT photocathode. The S13 light guides will be made up of eight segments, each 12.1 cm 2 in cross section. These will be formed to map the scintillator exit area onto 44.2 cm 2 at the end of the light guide. The light guides will be coupled to the PMTs by a machined UVT transition piece which begins as a 44.2 cm 2 rectangle and progressively tapers into a 4.6 cm diameter cylinder. Similar tapered transitions are used in the IMAX and ISOMAX TOF systems. The S3 light guides will be simple \sh tails" whose main purpose is the extend beyond the calorimeter volume and smooth out the position dependence of the S3 response Magnetic shields The R2083 PMTs chosen as the baseline for PAMELA require very low magnetic elds in the region between the photocathode and the rst dynode. Thus, some magnetic shielding will be required. However, if the magnet itself is shielded, the magnetic elds at the TOF locations will be low enough that they can be accommodated with very light shields. These will be made up of two layers of 0.1 mm thick \mu{metal", extending about 5 cm in front of the photocathode and 10 cm behind it. The magnetic shields will also serve as electrostatic barriers to protect the photocathodes (operated at negative high voltage) from eld emission damage. The full shield for an R2083 will weigh no more than about 40 g. The R5900s used on S3 will not require shielding Electronics The discriminators and time digitizers used in PAMELA must be capable of time resolutions better than 100 ps. This is well within the capability of current electronics in use at accelerators. However, these electronics generally have not been developed with the requirements of space ight in mind. This is particularly true with regard to power consumption. A typical discriminator and TDC packaged in a CAMAC format consumes about 3.5 W/channel. This can be reduced to about 1 W/channel by the elimination of CAMAC communications and ECL interconnections. Work is under way atgsfc and to build low power TOF electronics for future balloon and satellite based instruments. In addition, low power TDCs with 1 ns resolution are under development at the University of Siegen. This work is particularly directed toward the requirements of the balloon{borne ISOMAX experiment, which is currently under construction. In its long duration conguration, ISOMAX will require electronics whose performance and power consumption are quite suitable to the requirements of PAMELA. At this time the power consumption estimate for the new electronics is between 0.4 and 0.7 W/channel for the discriminator, TDC and ADC. The low power electronics development is at an early stage and but it is expected that the PAMELA TOF electronics will derive considerable benet from these developments. The PAMELA TOF will also require PMT bases which use very little power compared to conventional resistive divider chains of similar performance. Both a Cockroft{Walton voltage multiplier and an active base using a resistive voltage divider with FET current{followers for the last few dynodes [4, 11] will be evaluated. 73

76 Design of the PMT base to preserve the high frequency (fast rise and fall) performance of the PMT is critical to the performance of a TOF system. As a result, most of the TOF systems developed to date have used resistive voltage dividers (often with a current amplier or \booster" supply attached to the last few dynodes to improve high rate performance). Based on this experience, the baseline for the PAMELA TOF system will be a low{current voltage divider made up of high reliability metal{lm resistors. In order to provide high{rate capability, the dividers will use FET current followers on the last three dynodes. The nominal divider current will be 100 A, although it may be possible to reduce this to 50 A. Taking power supply eciencies into account, the bases will consume between 0.23 and 0.45 W each. The S3 detector will require only ADCs and PMT bases. These can be very low power consumption designs. The ADCs will require less than 0.05 W each and the PMT bases less than 0.07 W. Thus, the baseline (eight PMT) detector will require less than 1 W. With these bases and the low power electronics, the power consumption of the TOF system is expected to fall between 0.63 and 1.15 W/channel. For the baseline system with 26 PMT baseline TOF system this will result in a power consumption of between 16.4 and 30 W. In addition, on the order of 1 W will be required to interface the TOF to the instrument data system. Thus, the total power consumption of the scintillator system, including S3, is expected to be between 18.4 and 32 W Construction Each of the scintillator strips will rst be assembled as a complete, light{tight, unit with light guides and photomultiplier tubes. The light guides will be attached to the scintillators using a robust polyurathane{based optical glue (e.g. Hartel He{17017 as used in the SMILI, IMAX, and ISOMAX balloon instruments). The same glue will be used to attach the transition pieces to the light guides. The PMTs will be coupled to the transitions using a exible silicone (General Electric RTV{615) layer. This may be congured as a glue joint (as in IMAX and ISOMAX) or the PMT may be held using external spring pressure. The decision on which system to use will be made during the nal design phase. The scintillators will be wrapped in a thin layer of black paper to prevent adhesion and will be made light tight using a thin heat{shrunk plastic cover. The individual scintillator strips will be combined into layer and group subassemblies by attaching the strips to light{weight carbon{ber support substrates. These will serve both to provide support for the scintillators and to raise the resonant frequency of the assembly. Final support will be provided by attaching the subassemblies to the PAMELA instrument framework. So attached, the TOF system will be able to withstand the expected launch stresses. To help isolate the PMTs from launch stresses, the PMT mounting to the support substrates will be somewhat compliant. Together with the exible coupling to the scintillators, this will insure that the PMTs survive launch without damage Weight The estimated weight of the baseline TOF system, including a 10% allowance for structure is 27.5 kg. A similar estimate for S3 is 2.4 kg. The full TOF electronics system should weigh less than 3 kg. Thus, the full system weight is expected to be less than 33 kg Performance The average amount of light collected for a given energy deposit in the PAMELA scintillators will be about 30 to 40 percent higher than in IMAX due to the short length of the strips and the improved match between the end area of the scintillator and the photocathode area of the PMTs (in IMAX, the photocathode area was only 83% of the scintillator exit area). As a result, the timing resolution between any two PAMELA TOF strips in the top and bottom groups should exceed the measured IMAX performance of 122 ps for relativistic singly{charged particles. It is expected that the timing resolution will, in fact, approach 100 ps. As noted above, the use of two layers in the top and bottom groups results in a factor of 1.41 improvement in the overall timing. Thus, the full instrument timing resolution should be considerably better than 100 ps. While the timing resolution of the S13 layer will be comparable to that of the other layers, it will only improve the velocity resolution of the TOF system slightly due to the reduced particle ight path of about 50 cm between S13 and the bottom TOF group compared to the full TOF ight path of 88 cm PAMELA TOF Summary Dimensions: { S11, S12: active area 4540 cm 2 74

77 { S13, S21, and S22 active area: 1715 cm 2 { S3 active area: 3632 cm 2 { Scintillator thickness: 1 cm each layer, 2 cm each group of two layers. Group (S11+S12, S21+S22) thickness: 3 cm with support structure and wrapping materials Weight: 33 kg Power consumption: 18.4 to 32 W Time resolution: <120 ps per pair of strips, <100 ps between groups. Figure 7.1: PAMELA anticoincidence scheme 7.4 The anticoincidence counters A set of scintillation counters protects the tracker system from the contamination of particles coming from the side of the magnet (CAS counters) or from the top, but outside of the geometrical acceptance of the telescope (CAT counters). Their fast signals are read out in correspondence of each registered event and sent to Earth with all the other informations. The amplitude and time of possible on time signals will help in the understanding the pattern of the registered event. There are two set of such counters: 4 counters covering the top edge of the magnet constitute the CAT system, and 4 lateral hodoscopes composed each by 5 counters and covering the sides of the magnet constitute the CAS system (see Fig. 7.1). The 4 top counters are individually wrapped in alluminized mylar and assembled in a unique light tight box. The box is xed to the top shield of the magnet. The 5 counters of three of the four sides are individually wrapped in the alluminized mylar and xed in light tight boxes (one for each side). Each box is xed to the respective side of the lateral magnetic shield. On the forth side, the 5 counters are closed in separate light tight boxes in order to keep free 10 mm between them for give access to the planes of the tracker and their electronics (see g. 7.1). For all the anticoincidence counters, where the limited photocatode area is not a drawback, the Hamamatsu R5900 photomultipliers (two for each counter) have been chosen as a baseline PAMELA anticoincidence Summary Dimensions: 75

78 { CAT active area: (216 cm 2 )4 { CAS active area: (309 cm 2 )15+(308 cm 2 )5 { Scintillator thickness: 1 cm each Weight (including the mechanical structure): 13 kg Power consumption: 5 W 76

79 Bibliography [1] Ahmad, S., et al.: 1993, Nucl. Inst. and Meth. A330, 416. [2] Albergo, S., et al.: 1996, LBNL Technical Report [3] Beatty, J.J., et al.: Ap. J. 413, 268. [4] Kerns, C.R.: 1977 IEEE Trans. Nucl. Sci. NS{24, 353. [5] Mitchell, J.W., et al.: 1993a, Proc. 23rd ICRC, 1, 519. [6] Mitchell, J.W., et al.: 1993b, Proc. 23rd ICRC, 2, 627. [7] Mitchell, J.W., et al.: 1996, Phys. Rev. Lett. 76, [8] Streitmatter, R.E., et al.: 1989 Adv. Space Res. 9, (12)65. [9] Streitmatter, R.E., et al.: 1993 Proc. 23rd ICRC, 2, 623. [10] Sugitate, T., et al.: 1986 Nucl. Inst. and Meth. A249, 354. [11] Takeuchi, S., and Nagai, T.: 1985 IEEE Trans. Nucl. Sci. NS{32, 78. [12] Yamamoto, A, et al.: 1994 Proc. of the 5th BESS workshop, KEK. G4{11 77

80 78

81 Chapter 8 Transition Radiation Detector 8.1 Introduction In high energy or astroparticle physics the observation of transition radiation (TR) can provide valuable non{ destructive information for particle identication, in addition to calorimetric measurements. Fig. 8.1 shows the place of TRDs on the energy scale among other detectors for particle discrimination. The rejection factor denition for two kinds of particles is R =e = =e e = 90% (8.1) where ;e denotes the detection eciencies for pions and electrons respectively. In upper part of the gure it is shown the detector lengths for various detection techniques to discriminate various particles at 1% level of rejection factor. The lower part shows the threshold character of transition radiation, which starts below 1 GeV/c momentum and saturates slightly beyond that value. This radiation onset is typical for relatively small length TRDs due to the limitation of the gap between the foils of the radiators (optimal value about 200 m). The number of TR photons per cm, called \radiation ability", is extremely small (0.1{0.2); therefore in order to get rejection factors below 1% the detector length should be of the order of 50 cm, as shown in Fig According to this compilation the prototype used to design the TRD for TS93 balloon borne experiment [1] gives very competitive results. For the same TRD, in a shorter conguration, namely four radiator/chamber sets, of 26 cm total length, we got 3%. In this proposal we discuss the design of a TRD of equivalent length (about 28 cm), in order to get a similar result with a more reliable X{ray detector, robust and suitable for a long duration ight such as a RIM satellite mission. 8.2 Radiator optimization As we said before the identication power of TRDs depends very strongly on the radiation ability: for example a 15% change of this parameter gives a factor 2 in rejection power according to Monte Carlo calculations, supported by several tests. For a radiator completely transparent to its own radiation the photons/cm result to be just 1, because the emission probability per interface is of the order of =1/137 and 1 cm radiator is based on about 50 foils 200 m spaced apart as we said before. We can conclude that in all the TRDs of Fig. 8.2 (0.1{0.2 radiation ability declared) about 90%{80% of the photons are absorbed in the radiator. The simplest way to increase the radiation ability is to sample as much \nely" as possible the radiator/detector sets. The main parameters to be controlled in this procedure are the total number of electronic channels and the feasibility of a ne grained X{ray detector. Many attempts have been done according to this idea trying to get 0.4 photons/cm with realistic radiators, but nothing successful has been reported so far. It can be possible with particular carbon bers of suitable thickness and spacing to improve the radiation ability from 0.24 X{ray/cm (TS93 prototype in the best conditions) to 0.3 or even to 0.4: it is important to study in this case the ber segment length and its bulk density more carefully than was done before. We have xed the ber length to 7 mm and radiator density to 60 g/l. It is worthwhile to point out that with our radiator setting we already got the best radiation ability achieved so far, without any particular design or calculations. 79

82 Figure 8.1: rejection factors of detectors for particle identication 8.3 Straw tube detector for transition radiation When reducing the sampling of radiator/detector particular attention should be paid to the detector material and thickness. Typical detectors for TRDs are gas chambers in order to reduce the ionization energy loss from the primary charged particle; in order to detect eciently X rays of about 10 KeV energy, xenon gas (about 10 mm thickness) quenched with methane or carbon dioxide must be used. Any other solution such as scintillators or solid state detectors have been always disregarded because of the unfavourable ratio TR signal to particle ionization background. In our case we need a gaseous detector which should be robust, reliable and possibly works in sealed mode to be used without frequent gas changes for three years. The ideal detector appears to be the straw tube. It is basically an aluminized plastic proportional tube of 4 mm to a few cm diameter. It is made usually wrapping around the axis two strips of aluminized mylar or polycarbonate glued together to form a cylinder with a wall 25{100 m thick. They can be operated even to 1-4 Atm and yield resolutions of 25{100 m. The advantages of a straw tube detector when compared to a MWPC are: 1. the straws are inexpensive, robust and simple to mount; 2. the damage and possible down time caused by wire breakage is minimal since the broken wire is isolated in the tube cell; 3. the eects of signal cross talk are minimized as the straw cathode provides a complete ground shields between nearby wires; 4. the problems of electrostatic alignment distortions are minimal when the anode is kept centered in the straw; 5. the wall thickness allows overpressures of 4 atm, so that it is possible with accurate design of the end plug and assembling procedures to keep gas tightness for long periods. 80

83 Figure 8.2: rejection factors of TRDs versus length Cathode materials We have investigated the gas permeabilities of various plastic materials, and we have noticed that except for PET (polyethylene{terephthalate), most of them do not exhibit a quite reduced diusion rate for common gases. The Kapton (polyimide) permeability for molecular oxygen through a foil 25m thick, 100 square inches of surface area, is 25 cm 3 per 24 hours at 1 atm. This value is almost comparable to that of PET and is twelve times better than polycarbonate which is commonly used for straw tubes. Similar values are reported for the other common gases, but unfortunately noble gases permeabilities are not quoted [2]. Only recently very good reports have been made on 50 m thick Kapton straw tubes [3] for TRD applications, both from the mechanical and from the radiation hardness point of view. The other attractive feature of Kapton is the higher transparency to soft transition radiation photons with respect to Mylar, which is about one third worse, and the very low radiation and interaction length. For these reasons we chose Kapton tubes of even thinner cathode walls: we were able to purchase 30 m thick, 4 mm diameter straw tubes [4] which is the minimum provided or reported so far. The Kapton straws are internally coated with a copper layer which has an ohmic resistance of 0.15 /sq. The copper work function (4.65 ev) is slightly better than that of aluminum (4.28 ev). In addition its conductivity is superior to that of aluminum. However the unavoidable oxide layer that forms on the metallic surface reduces by a factor 10 8 the conductivity respect to pure metal, although the resulting CuO conductivity is10 12 higher than that of Al 2 O Anode wires Since our straw tubes are shorter than 50 cm, and will be operated with gas mixtures based on xenon and CO 2, we have preferred to use thin anode wires in order to work with reduced high voltages and to limit the risk of discharge between the tube cathodes and the anode feedthroughs. For mechanical safety reasons we preferred not to go below 25m thickness. We tested dierent counters equipped with tungsten wires stretched to tensions ranging from 50 to 70 g. Due to the short length of these prototypes we did not need any intermediate spacer to avoid wire electrical instabilities Straw tube electrical connections The tube end plug consists of three dierent coaxial elements inserted one inside the other (Fig. 8.3). The rst element (ground plug) is made by precision machined brass, is equipped with a transversal gas brass pipe and is gold plated. It is inserted inside each straws end and provides both gas circulation inside the tube and cathode electrical connection to ground. The brass gas pipe connects the ground plug to the gas manifold built inside the end plate. 81

84 Figure 8.3: straw tube end{plug section A vetronite insulating plug is inserted inside the ground plug, and it decouples from ground the third element, namely the inner gold plated brass pin which keeps the anode wire. For this purpose the pin has an axial hole, 200 m diameter [5] where the tungsten wire is inserted and soldered. With 70 g nominal tension all the wires work properly and we did not found any broken one End plate In our design we decided to group the straw tubes as a modular unit consisting of two layers each of 16 adjacent tubes that are nally stacked to form a double layer close pack conguration (Fig. 8.4). In this way we ensure mechanical stability and consistency to the tubes. The straws, after the ground plugs insertion, are mounted on the aluminum end plates (Fig. 8.5), previously secured on a precision transfer frame. Only after this procedure the anode wire can be inserted, stretched and soldered, in order to be sure that the resulting mechanical stress is distributed mainly over the transfer frame, that will be removed right before the denitive mounting of the complete module on the nal frame. After these operations the twolayers are coupled face to face, the two transfer frames are connected together, and glued with liquid epoxy. In this way the counter is gas sealed. After this procedure the module can be pressurize up to 4 atm absolute. Generally the modules exhibit gas leaks for argon based mixtures of about 3{410,3 Torrl/sec at 1 atm overpressure, which is good agreement with the permeabilities quoted for common gases (noble gases are not quoted on the reference). The modules can be nally mounted on the proper frame, side by side, to form a chamber, with very reduced size inecient zones between adjacent ones (Fig. 8.4). The gas is distributed connecting the modules in series, by means of Tygon tubes segments inserted for 1 cm on the aluminum pipes of the end plates. 8.4 Tests on straw tubes prototypes with radioactive source We tested some straw chamber prototypes lled by Xe/CO 2 gas mixtures using photons of 5.9 KeV from a 55 Fe source to investigate both the gas gain and the pulse height behaviour versus high voltage, since we intend to detect transition radiation of the same energy. We have now started to carry out longer term tests, namely runs of a few months duration at high irradiation doses to study the chamber performances in radiation conditions equivalent to those expected on our satellite orbit. In Fig. 8.6 we show the gas gain curves for two gas mixtures. The high voltage needed to run in a moderate gain region, i.e. of the order of 10 4, is about 1400 V. The electronic channel used was based on a common base input transistor, followed by a commercial amplier NE 592 and a buer LH0002 to drive the signal to a discriminator or to an ADC on a 50 ohm coaxial cable. 8.5 Straw chamber tests in vacuum on a beam line In order to test our prototypes performances in vacuum and to study their eciency, operation stability, radiation damage and ageing we built 1000 straw channels, of 20 cm active length. We installed them inside the vacuum chamber (close to the target) of the experiment E864 at B.N.L. 82

85 Figure 8.4: drawing of a pair of adjacent straw tube modules Figure 8.5: artist view of a mono{layer end{plate 83

86 The straw modules are arranged in three parallel planes at dierent orientations, namely 0, +20, -20 with respect to the vertical direction to allow tracking capability. The complete detector has dimensions of about m 2. The overall interaction and radiation length are ,3 I and 10,3 X 0 respectively. Figure 8.6: pulse heights (right scale) Xe/CO 2 gas mixtures: dots Xe(80%); squares Xe(70%). The tted lines are drawn just to guide the eyes. The modules of each plane are mounted on an aluminum rectangular frame, 1 cm tick, that has been xed vertically on the upper side of a horizontal aluminum ange. This ange is crossed by three parallel printed board plates, for all its length, that bring the electrical connections to the straw planes. The bottom pins of the straw modules are connected to the boards through Teon insulated thin cables. Each printed board is equipped with blocking capacitors and H.V. supply resistors and consists of nine layers to provide high voltage independent connection to each straw module, signal connections to each module layer, and appropriate shielding layers between signal traces and H.V. buses. On the outer part of the ange (outside the vacuum) each board is equipped with edge connectors to plug in PCOS IV LeCroy read{out cards, borrowed by the E864 collaboration. Particular care was paid to the proper electrical insulation of any exposed H.V. element inside the vacuum to avoid glow discharges. The detector was fed at a few cm 3 /min with an Ar(90%){CO 2 (9%){CH 4 (1%) mixture of the experiment, and was operated at 1150 V. The overall gas leak was monitored carefully before the installation inside the vacuum chamber and was about 710,3 Torrl/s at 1 atm overpressure, namely 0.5 cm 3 /min for a 3.3 l volume detector. The detector has worked for about one month during the fall '95 in the vacuum chamber at 20 mbar pressure and presumably has been exposed to about 10 krad of radiation without any appreciable loss of vacuum tightness. The experiment run up to particle/spill with a Au 10% target. The straw detector has been exposed to most of charged secondary particles, low mass ions included. The data analysis is in progress to establish the operating stability of materials used (Kapton, epoxies, etc.) in vacuum and at the radiation level above mentioned by studying the homogeneity of the eciency and tracking capability during the exposure to the beam. Interesting results for the TRD are expected by the data collected with low mass ions. 8.6 TRD design On the basis of the straw module design we have decided to built the TRD for PAMELA using a compact assembly structure as shown in Fig

87 Figure 8.7: TRD structure: cross sectional view It consists of 12 layers of straw modules interleaved by carbon ber radiators which ll completely the gaps between the tubes. The total number of modules is 60. They are arranged as three superlayers (of four layers each) of degrading extension (from 30 to 20 cm length) to form a truncated pyramid structure. The radiator thickness (about 15 mm) is determined by the space left when assembling the straw module layers on sti carbon ber square frames 3 mm thick. The carbon ber density has been xed to 60 g/l. All the frames are stiened by precision steel rods running through them. This structure is placed inside an aluminum box (Fig. 8.8) which is sealed and then evacuated. During the ight it will be in communication by a remote controlled valve with the vacuum outside the payload, in order to vent out at xed time intervals the gas mixture leaking from the straw tubes. Figure 8.8: vacuum box artistic view; from the front plate the HV{signal connectors holes are shown We think that because of this leak due to the straw wall permeability a storage gas vessel of about 1.5{2 cubic meters (at atmospheric pressure) will ensure three years of TRD operation. The aluminum box is equipped with a top and bottom thin steel window to allow the particles to pass with minimum energy absorption or scattering and a front ange that brings the vacuum high voltage connectors and signal cables outside to the electronics boxes housed at the two sides. 85

88 In these boxes are located the blocking capacitors, supply resistors, and read out electronics. 8.7 Straw tube TRD expected performances According to the above considerations the TRD can discriminate electrons/positrons from protons/antiprotons starting eciently from 1 GeV/c momentum. The total number of channels is The radiator should yield, according to the radiation abilityachieved, 5{6.4 TR photons (0.3{0.4 radiation ability). The gas total thickness in the close pack conguration is 2mm 12 = 7:5 cm, and yields 0.11 /cm 7.5 cm=0.9 {rays (above 5 KeV) which constitute the unwanted particle background. This background is slightly less than that exhibited by the TS93 prototype TRD. According to Monte Carlo (well tested) calculations such a TRD with only 28 cm length would give from 5% to 2% rejection factor starting at a momentum of less then 1 GeV/c: a similar detector based on straw tubes [6] yields similar results but with dierent radiator materials and design. However it should be stressed that a length reduction of say 20% (20 cm length) would worsen this result to 15%{6% rejection factor. This TRD could have also the additional advantage to measure the charge of cosmic ray nuclei, since it is based on a gaseous detector working in proportional mode with a resolution better than any scintillator space detector used so far due to the ne sampling of the tube layers that will yield 12 independent measurements proportional to Z TRD electronics The processing of signals from the detector based on cluster counting technique ensures a stable rejection factor with respect to signal amplitude variations. This techniques has already been successfully used in the WiZard experiment TS93. To meet the stringent power limits imposed in PAMELA, a new, simple and low{power front{ end electronics has been devised. It is based on a semicustom analog array, available from Thomson CSF, in which bipolar transistors having high gain{bandwidth product at low current allow to match the short shaping time required by cluster counting with low power consumption. Being the signal charge in the order of 3{510 5 electrons, the noise is not a severe problem for the design. First results of simulations show that the digital section of the front{end could be realized on a Actel eld{programmable{gate{array, already used satisfactorily in WiZard experiment. The low event rate will allow a clock frequency of 1 MHz. In these conditions, the power consumed by the digital section will be contained within 5 mw/channel. Thus, the total power consumption of the detector will be 20 mw/channel. The HV of the straw tubes will add a negligible contribution. A block schematic of the TRD electronics is shown in Fig. 8.9 The current signal coming from each straw tube is amplied by a current{sensitive amplier and discriminated to select cluster signals. The delay is needed to wait the arrival of rst{level trigger. The cluster pulses are then stored in a counter. When the reading process starts, the data are daisy{chained towards a buer memory. Zero skipping is made on{the{y. 8.9 TRD data acquisition and local processing Fig shows the data acquisition system of the detector. The system is centered on a microcontroller (we are considering Motorola controller series 68HCCXX). In order to speedup the data acquisition, the cluster counters are grouped into four daisy chains driven in parallel. With a 1 MHz clock frequency, the daq process will be completed within 0.4 msec. The main tasks done by the processor will be: collection of raw data from cluster counters (3 kbits/event), skipping simultaneously the zeros; rejection of spurious hits; addition of proper wire address to accepted data; implementation of any other data reduction procedure; storage of reduced data, waiting for an interrupt from the main processor. According to the amount of data reduction performed by the local processor, communication with the main processor will take place either through a serial link or a dual{port memory 86

89 TRD ELECTRONICS FOR PAMELA VTH GATE TEST 8mW AMP 5mW COMP DELAY A B A Q B AND2 Q OR2 2-BIT CNTR 1 SELECT 2 ANALOG ARRAY (12 CH.) FPGA 1920 <5mW/CH SELECT CLK ADDR CNTR 11 ADD A B OR2 A B ZERO SUPPR. Q Q AND2 W MEMORY 2 DATA CHANNELS -CLOCK 1 MHZ -READOUT 400 USEC LOCAL PROCESS. MAIN PROC. Figure 8.9: scheme of the TRD electronic. 87

90 SELECT SHIFT FROM IRQ MUX TRD DAQ SHIFT REG4 SHIFT REG1 STRAW 1500 CH THRESH 2-BIT CNTR 2-BIT CNTR CLEAR up FROM TRIGG MAIN BUS DAT DP MEM ADD Figure 8.10: data acquisition system of the TR. 88

91 8.10 ADC based read{out A backup DAQ design in case of choice of linear analog read out solution, instead of cluster counting solution, has been studied optimizing power consumption and reliability of the apparatus. The processing of the signals from the detector is based on analogic technique. The front-end is based on the VIKING chip. This one consists of 128 charge sensitive preampliers and 128 CR{RC shapers followed by a sample and hold circuitry, input and output multiplexing and an output buer. Fig shows the data acquisition system of the detector. The system is based on a microcontroller, HITACHI SH7032, which has been selected for its exibility. The system is composed of 3 HITACHI SH7032, 3 ADC COMLINEAR CLC945 and 15 VIKING chip. This conguration has been chosen because the straw tube system is composed of 3 groups of 4 ats that our system allows to read independently. The output of the straw tube is connected to the input of VIKING whose analog output is coupled to the input of the ADC. This one converts its analogic input in digital output signal which is read and stored by the microcontroller HITACHI. After all information is recorded in each microcontroller, each one, under the direction of the MAIN, sends all data to the MAIN by synchronous serial link at 5 Mb/s rate. In the proposed readout scheme the overall time neded for the conversion and the transmission is 3.46 ms. The power consumption are: 1.5 W for the three controllers, 0.23 W for the three ADC and 1.73 W for the VIKING, corresponding to 1920 channels with a consumption of 0.9 mw/ch. The timing take into account the capability of the controllers to work under the DMAC (Direct Memory Access Controller) control. Using external and on-chip memory transfer, we can handle external address and data buses to store our data on an external memory. This buses are 22 and 16 bit wide respectively. The use of an external buer line SN74ACT7802 allows a total time for the acquisition and the transmission of s. 89

92 Figure 8.11: data acquisition system in case of signal amplitude read out. 90

93 Bibliography [1] E.Barbarito et al., Nucl. Instr. Meth A 313 (1992) 295. [2] Modern Plastic Encyclopedia, Mc Graw Hill,N.Y. [3] V.Bondarenko et al. Nucl. Inst. and Meth. A327 (1993) 386. [4] Straw tubes are provided by LAMINA Dielectrics Ltd, Billingshurt, Sussex RH 149SG, England. [5] Anode wire pins are provided by Multicontact AG, Basel, Switzerland [6] Fabjan, Dolgoshein, J.T.Shank et al., Nucl. Inst. and Meth. A 309(1991)

94 92

95 Chapter 9 The magnetic system The magnet system is the core of the apparatus and special studies have being carried out in order to optimize design and performance. The VNIIEM factory has experience and suitable technologies to design and build permanent magnetic system for space application using materials with large magnetic remanence. The VNIIEM has already successfully used this kind of magnetic system in space and therefore we are condent of the reliability and the space qualication of these techniques. The main constraint in the PAMELA magnet design is the mass limit: 140 Kg including any shield to enclose the magnetic eld or to fasten the pieces. The mass limitation prevents the use of a superconducting coil. The results of evaluations performed by VNIIEM gave a maximum lifetime of few months. Thus the only feasible solution is a magnetic system built with permanent magnets. 9.1 Magnetic materials The use of permanent magnetic materials with high magnetic remanence permits to obtaining high magnetic elds with a low weight. The choice falls on rare earth sintered materials, like Nd{Fe{B, in which the magnetic particles are fused together in a pressing and sintering process. The residual magnetic induction (B r ) of these magnetic materials ranges between 1 and 1.3 T, depending on the working process quality. In particular, the quality of such magnets depends on precise adherence during the technological process, on alloy composition, on its dispersion, on temperature and time modes of treatment and moulding as well. At present, Nd{Fe{B magnetic materials, with a residual magnetic induction ranging between 1 and 1.2 T, are available at VNIIEM. The main characteristics of these materials, manufactured by VNIIEM, are summarized in Table 9.1. A proportional increase of the eld in the spectrometer should be expected using alloys with higher B r. Residual Coercive BH B r H c Tmax Alloy induction force B r (T) H c (ka/m) (kj/m) (%/k) (%/k) ( C) n n n.3 1.0{ { Table 9.1: magnetic properties of VNIIEM magnets from Nd{Fe{B alloys. Unfortunately, a single permanent magnet has limited geometrical dimensions (not more than mm 3 ) because of a complicated manufacturing procedure that determines diculties in building large size magnetic pieces. Nevertheless, gluing together several pieces of sintered Nd{Fe{B blocks and cutting them in suitable way, it is possible to obtain large magnetic pieces of any shape. 9.2 General consideration on PAMELA magnet conguration The magnetic bore cross section is xed by the silicon sensor dimension used. Each detector plane is composed of 6 silicon crystal of cm 2 for a total area of 1416 cm 2. With this sensitive area the exposure geometrical 93

96 factor for the cosmic ray measurement is dependent on the magnet height (see Fig. 9.1). For instance, with a magnet height of 45 cm, the geometrical factor is about 25 cm 2 sr. Geometrical factor [cm 2 sr] height [cm] Figure 9.1: Geometrical factor as function of the magnetic height with an area of 1416 cm 2. The magnetic system conguration, chosen to optimize the intensity and the uniformity of the magnetic eld inside the spectrometer, is the \yokeless model". The basic idea of this model is that, in a two dimensional model, it is possible to generate a uniform magnetic eld in any prismatic cavity using uniformly magnetized blocks only, without any yoke. As example, in Fig. 9.2 is shown the scheme of the yokeless magnet in the case of a square cavity and a magnetic eld inside of 0.293B r, where B r is the residual induction of the magnetic blocks used. We should point out that a shield is necessary to fasten or glue the magnetic blocks even though this magnetic design does not require an iron yoke. The theoretical eld value obtained from the the two dimensional conguration (or, equivalently, from a magnet of innite height) is modied when we consider a real magnet system with a nite height. The eld in the cavity is reduced and the magnetic map is not uniform, with a further reduction at the edge. The general technical requirement for the PAMELA magnetic system are: 1. The total weight must be less than 140 kg. 2. The average eld inside the spectrometer must be 0.35 T. 3. No strong inhomogeneities. Less than 10% in each cross section plane of the cavity and less than 50% in the whole cavity. 4. External eld (leakage eld) not more than 10 G at the distance of 50 cm from the magnetic system centre 5. Geometrical factor not less than 20 cm 2 sr. 6. The magnetic system structure should sustain the overload of 1{10 g with frequency of 10{2000 Hz and 40 g during 0.5{6 ms of shock duration. 7. Feasibility of 6 sets insertion of silicon planes with dimension of mm 2 and their easy withdrawal from the magnetic system should be envisaged. 8. The system should be operational in the ambient temperature range (dry nitrogen) from 5 Cto40 C. 94

97 B Figure 9.2: Yokeless magnet system with a eld inside the cavity of 0.293B r (B r is the residual induction of the magnetic block used). Due to the suitable choice of the magnetic piece shapes and the directions of the residual induction, the eld in the cavity results uniform. Calculations and theoretical studies on dierent systems, and technological considerations also, have shown that the yokeless model, like that shown in Fig. 9.2, is the best system which t the requirements 1, 2 and 3. This magnetic system has small leakage eld, which is important to minimize the eect of the magnetic eld on the near electronic devices. But the calculation show that, with a eld of 0.3{0.5 T in the gap, the leakage eld is 150{200 G at a distance of 50 cm from the magnetic system center. This problem is solved by the insertion of a thin ferromagnetic shield, externally to the magnetic pieces, and its weight has to be taken into account. 9.3 PAMELA Magnet design The inner section of the magnetic spectrometer is chosen to be mm 2 so that the silicon plane of mm 2 can be accommodate inside. The general schema of the magnet system is shown in Fig It consists of ve modules 8 cm height interleaved by six frame 8 mm height, in which the silicon sensors are accommodate. The total height of the spectrometer results 44 cm corresponding to a geometrical factor, at high energy, of 24.5 cm 2 sr. Each magnetic module is constituted of a case of non{magnetic light material (aluminium) in which the magnetized pieces, with B r =1.1 T, are glued. The horizontal cross section of one module is shown in Fig. 9.5, where the arrows indicate the directions of the magnetic inductions in a typical yokeless model conguration. In Fig. 9.6 is shown the vertical cross section of a magnetic module. It is visible the aluminium case in which the magnetic pieces are glued and the lateral holes on it, to x the modules each others and to the silicon frame. In order to reduce the leakage eld outside the spectrometer, it is foreseen a 2 mm thick of ferromagnetic screen put at 3 cm around the magnetic system, with two windows in correspondence of the top and the bottom magnet gaps. In the shield there are slots, 12 mm height, to permit the silicon planes insertion. The lateral view of the magnetic system, with included the screen, is shown in Fig The design of the PAMELA magnet system, based on permanent magnet, requires special attention of organizational, technical and manufacturing items: integration of separate magnets into modules; module mechanical treatments providing quite high level of precision and of treatment quality; 95

98 module magnetic property check; Figure 9.3: Scheme of the magnetic system. It is composed of ve modules of 80 mm height interleaved by 6 frame for the silicon detector. The gap between the magnetic modules is 8 mm height. It is visible also the external ferromagnetic screen. magnetic system integration of modules having large magnetic elds which cause appearance of attraction{ repulsion large forces between them; special tooling design and manufacturing out of non{magnetic materials for integration; special working places arrangement for magnetic system integration and testing; Based on these requirements, the following magnetic system manufacturing procedure is planed: 1. Permanent magnet blocks, based on Nd{Fe{B alloy,of mm 3 are manufactured and their properties are checked. Then, these blocks are subjected to nal treatment: grinding on a at grinder with the usage of special tooling to provide precise relative position of surfaces. 2. Magnet blocks are integrated and glued in a special device. The \Anaterm{107" anaerobic, shock{proof adhesive, is used for glueing. Adhesion of 25 MPa is provided. 3. Workpieces having the form of rectangular prisms are cut, in modules of the required form, on the \Opticut" electric erosion machine. As an example, in Fig. 9.7 a rectangular unit with dimensions of mm 3 is cut into two three{angle prisms used to form the yokeless magnet of Fig Modules are magnetized on a pulse installation using a special multi{turn inductor to provide total magnetization of modules. Module xed position in the installation should be provided by a special device. 5. Units structural components to provide strength and rigidity are made out of non{magnetic material using the method of mechanical treatment (planing, milling, drilling). 6. Modules are glued to the structural components and the units are integrated taking into account the mutual magnetic attraction{repulsion forces of modules. Depending on modules form, the following glue compositions are used: Anaterm{107 or K{300H providing the adhesion of 12.5 MPa. 96

99 Figure 9.4: Lateral view of the magnetic system with the ferromagnetic screen included. The slots which permit the introduction of the silicon planes are visible. 97

100 Figure 9.5: Horizontal cross section of one magnetic module. Magnetic inductions directions are visible in each magnetic piece. Figure 9.6: Vertical cross section of one magnetic module. 98

101 Figure 9.7: The gure shows schematically eleven blocks glued and cut to form two three{angle prisms for the construction of the yokeless magnet. 7. To protect magnets and to safeguard silicon devices against occasional leak of magnet fragments it is necessary to insert (glueing on the inner side of the magnetic system) protective spacers with the thickness of 0.2{0.5 mm made out of 50H alloy. 8. Storage and transportation of all magnetic system components at each stage of the manufacturing process should be performed in a specially designed crate for each component. Taking into account the structural peculiarities of the magnetic system as well as VNIIEM experience in magnetic system production for various electric devices (and space applications), the system is manufactured using the equipment available in NPP VNIIEM. 9.4 Magnetic map calculation It has been developed a simulation program to calculate the magnetic map (in three{dimensions) for any kind of magnetic system and ferromagnetic shield disposition. The calculation is based on the equivalence between the eect of the residual induction in a magnetic block and the eect of a \magnetic charge" on its surface. The numerical integration is performed by the integral equation method. This code has been used to optimize the magnetic system design. In Fig. 9.8 is shown the magnetic intensity along the magnet axis. The eld intensity in the center of the system is 0.38 T and at the top{bottom, where the eld is reduced due to the nite height of the magnet, the eld is 0.18 T. So, the eld variation along the axis is about 50%. The average eld in the whole cavity is 0.35 T. In the Fig. 9.9 is shown the eld intensity in the central cross section perpendicular to the magnet axis. The eld direction is along the Y-axis and in the gure is evident the good intensity homogeneity of the eld, except in the corner regions where the contribution at the total acceptance is very small. In order to obtain a small leakage magnetic eld, the eect of the ferromagnetic shield has been carefully studied. As an example, in the Fig is shown the eld intensity in the plane where the S21/S22 scintillator is located. This plane is near to the bottom of the magnetic system where the eld is large. It is important to estimate the eld intensity in this plane because of its inuence on the phototubes behaviour. In the upper part of the gure is shown the eld intensity in the case of ferromagnetic screen introduction and in the lower part the intensity in the case of no ferromagnetic screen. With the screen the eld intensity is well reduced and at 20 cm from the magnet axis the eld is smaller than 5 G. The conclusion of this study is that the eld eect of the PAMELA magnet system on the electronic operation on{board is not critical. 99

102 Magnetic field [T] Z [cm] Figure 9.8: Magnetic intensity along the magnet axis. Field (T) Y (cm) X (cm) Figure 9.9: Magnetic eld intensity in the horizontal central cross section. 100

103 1 Field (T) Y (cm) with screen X (cm) 1 Field (T) Y (cm) without screen X (cm) Figure 9.10: Magnetic eld in the scintillator plane S21/S22 with screen (upper gure) and without screen (lower gure). 101

104 9.5 Test and space qualication The building of the magnetic system prototype has already done. It consists of one magnet module and four simulators, prepared with non{magnetic material, according to the mass and dimensions of the working module. The modules are assembled in one structure interleaved by the support frame of the silicon tracker planes. The whole system is closed in a ferromagnetic shield with the two windows in correspondence of the magnetic gaps. This prototype has been useful to elaborate the technological construction procedure and to test the magnetic spectrometer performances. In the next future the following tests will be performed: Magnetic test. calculation. The measured magnetic eld will be compared with the one expected by theoretical Temperature test. The magnetic eld map will be measured in dierent temperature condition to evaluate the eect of the temperature variation. Duration test. The measurement will be performed at intervals of one or two months to control the degradation in a long period of the magnetic materials. Mechanical test. The mechanical shock and vibrations during the launch will be simulated to study the eects on the mechanical system and on the magnetic eld intensity. Spectrometer test. Assembling few silicon plane prototypes it will be possible to test the spectrometer performance using cosmic ray muons. 9.6 Time schedule and plane of the work The Flight Model (FM) manufacturing procedure of the PAMELA magnet system could be summirized in the following steps: Manufacturing of structure elements Manufacturing of magnets{bars the FM magnetic system Manufacturing of structure elements of the FM magnetic system Assembling of the FM magnetic system Carrying out of magnetic system test A conservative total time evaluation for the PAMELA magnet system delivery is one year including the Nd{Fe{B material providing, the manufacturing itself, the test for space qualication and the transport. 102

105 Chapter 10 The silicon tracker 10.1 Experience with Silicon Detectors Since 1991, the L3 Florence group has been involved in the design and construction of the L3 Silicon Microvertex Detector (SMD) [1, 2], which consisted in 35 cm long very transparent (4:10,3 X 0 ) ladders, supported by avery light carbon ber structure, using double sided 300 m thick 74 cm 2 area silicon sensors, designed at INFN{ Pisa and processed at CSEM, Neuch^atel [3]. The readout pitch of these sensors were 50m on p{side (bending plane side) and 150{200 m on n{side (non bending plane side). Results from exposure to 50 GeV/c pion beam in 1991 show a spatial resolution of 7m on bending plane and 14.7m on the non-bending plane, respectively [4]; these results are conrmed from the actual running experience during the period of LEP data taking from 1992 to In particular, the Florence group was responsible for the design of the SMD readout electronics, and introduced the novel concept of optical decoupling to read out both side of the detectors [5, 6]. Moreover, in the last few years INFN did a considerable eort to realize in Florence a fully equipped laboratory for the development of silicon detectors, and an important group of people is involved in R&D activities in this eld, related both to the development of new detector's techniques and to the associated readout electronics. As a consequence of this large investment, the Florence group has today the facilities and know{how necessary to build the PAMELA silicon tracker General requirements and design goals of the silicon tracker The PAMELA tracker will consist of 6 equally spaced planes of silicon detectors, which will then be inserted into the mm 3 magnet free bore of corresponding Geometrical Factor GF=25 cm 2 sr. The tracker aims to fulll the two major following goals: 1. Measurement of the momentum vector of traversing charged particle including the determination of its charge sign, hence giving a precise measurement of the rigidity (momentum per unit charge) up to a Maximum Detectable Rigidity (MDR) of the order of 400 GV/c and allowing a sensitivity on antihelium/helium ratio better than 10,7. 2. Separate Z=1 particles from that of Z=2 using the de/dx capability of each plane (and possibly, the determination of the absolute charge for light nuclei up to Z=4). In the following section we will discuss the results of Montecarlo simulations carried out to establish the best design conguration to fulll the physics goals of the experiment. The main parameters of the PAMELA silicon spectrometer will also be summarized Optimization of the geometrical design The optimization of the tracker design was obtained through a detailed Montecarlo study of the detector with high statistics, using about 10 7 events for each set of parameters. The simulation included the motion of particles in the magnetic eld, the nuclear and Coulomb scattering processes in the tracking system and a procedure to measure the magnetic deection using a 2 minimization method. Each detection plane corresponds to approximately 3:2 10,3 X 0, and is essentially composed only by the silicon sensors, thanks to the mechanical structure realized by means of very rigid carbon ber stiners glued only on the lateral sides of the sensors, to minimize the total radiation length of each plane. The main parameters of the spectrometer including those used in the simulations are summarized in Table

106 Main Parameters of PAMELA Silicon Tracker Number of Planes 6 Geometric Factor 25 cm 2 sr Spatial Res.(Bending Plane) 7m Spatial Res.(Non Bending Plane) 15m Magnetic Field Strength 0.35 T Total Number of Channels 36,864 Table 10.1: Main parameters of PAMELA silicon tracker. The simulations were performed for several geometrical congurations varying also the number of the detection planes and the distribution of the planes inside the magnet free bore. It is evident that the maximum kinetic energy that we can reach for the antiproton ux measurement depends on the antiproton/proton ratio and on the acceptable background level. The main source of background at high energy (30 GeV) is due to the proton \spillover" eect, that is the wrong determination of the charge sign of traversing particle due to the scattering inside the spectrometer combined with the nite resolution of the silicon tracker planes. The number of detector planes should be as small as possible in order to reduce the scattering inside the spectrometer. To obtain a curvature measurement by performing also the pattern recognition, the minimum number of planes needed is 4. The simulations have shown that a 6 planes conguration gives the best compromise between the number of planes and the need of a good track eciency. To determine the best geometrical plane conguration inside the magnet cavity we have rst arranged 6 planes in three double planes (pairs) placed at the top, at the middle and at the bottom of the magnetic bore (over 45 cm total height of the magnet). Then we have varied the distance z between planes of each pair from 1.2 cm to 9 cm (9 cm corresponds to an equidistance between the planes). Fig shows the 2 track reconstruction distributions for \all events" and for \spillover events" for z=1.2 cm and for z=9 cm. From this gure we can see that the best separation is obtained with z=9 cm, hence it is possible to reject the maximum number of the proton spillover events. Therefore, in order to reach as high measurable energy for antiprotons as possible, it is important to have equidistant silicon planes even if there is a slight improvement in the spectrometer resolution for congurations with lower z. As a result of these simulations, we have therefore chosen the 6 equidistant planes for the nal geometrical conguration. Fig shows the spectrometer resolution in the magnetic deection for the nal conguration considering a tracking overall spatial resolution of 7 m in the bending direction and 15 m in the non bending one. The resulting maximum detectable rigidity is calculated to be 368 GV/c. Using Montecarlo, we have studied also the momentum resolution of tracker for the nal geometrical con- guration. The results then have been compared with those of the analytical calculations based on formulas given in reference [7] and were found to be in complete agreement. As it can be seen from Fig. 10.3, at very low energies (low ) the momentum resolution is dominated by the multiple scattering while for high momentum there is a rise of dp/p due to the nite spatial resolution of the tracker. For the momentum range of major interest, the spectrometer has a good momentum resolution, for instance for protons, 11% at 0.5 GeV/c, 6% at 5 GeV/c and about 29% at 100 GeV/c, and 6% for electrons at 0.5 GeV/c Energy Range for Antiproton Flux Measurements and for Antihelium Search As we already mentioned, the major limitation on the maximum measurable kinetic energy for antiproton ux measurements and for the antihelium search comes from the spillover eect. Fig shows the spillover background for various 2 selection cuts in the antiproton/proton dierential ux measurement as a function of the kinetic energy. The solid curve represents the expected ratio considering the secondary antiproton production in the interstellar gas of the Galaxy by the Modied Leaky Box Model [8]. The dashed curve indicates the expected antiproton/proton ratio by the Closed Galaxy Model [9]. The spillover backgrounds are given for no selection cut and with a selection eciency of 90%, 60% and 30% in the 2 track reconstruction selection cut. A conservative limit for the maximum measurable kinetic energy, Ek max, is dened when the spillover background is at least one order of magnitude less than the antiproton ux with a selection cut eciency of 60%. Therefore it is correct to say that the determination of a good antiproton energy spectrum is possible up to the spillover limit. As it can be seen from Fig. 10.4, for PAMELA, Ek max is 90 GeV for the Modied Leaky Box Model and 125 GeV for the Closed Galaxy Model. 104

107 arbitrary units arbitrary units χ 2 track reconstruction Log 10 (χ 2 ) χ 2 track reconstruction Log 10 (χ 2 ) Figure 10.1: Track reconstruction distributions of all events and of spillover events for z=1.2 cm and z=9 cm. 105

108 n. events d real -d measured [ c/gv ] Figure 10.2: Spectrometer's resolution in the magnetic deection for the nal geometrical conguration Delta P /P (%) 10 Protons Electrons P (GeV/c) Figure 10.3: Momentum resolution of PAMELA spectrometer for protons (crosses) and for electrons (stars). 106

109 Anti-p/p ratio Kinetic energy [ GeV ] Figure 10.4: The error due to the proton spillover in the antiproton/proton measurement compared with expectations of the Modied Leaky Box and Closed Galaxy Models. In Fig the expected antiproton counting rates for the Modied Leaky Box and for Closed Galaxy Models are plotted and compared with the spillover background (calculated for two 2 cut eciencies, 90% and 30%). For these calculations the exposure time of 1 year and the geometric factor of 25 cm 2 sr were assumed. In Fig is shown the spillover background in the antihelium/helium integral ux measurement as a function of the kinetic energy per nucleon. The curves represent the upper limit with a condence level of 90% in case of no antihelium observation in 3 years for a geometrical factor of 25 cm 2 sr. For the antihelium search the maximum measurable kinetic energy per nucleon is 30 GeV/nuc `State of the art' of silicon microstrip detectors In the last 10/15 years a huge eort has been realized by the particle physicists to develop new sophisticated silicon microstrip detectors; this eort has been done in strict collaboration with few relevant industries active in the eld [10, 3, 11, 12]. In this section we will present the `state of the art' of the technology used to build such kind of detectors, and we will present the sensors which are well suited for the Pamela tracker Double sided detectors The main progress has been the construction of double sided sensors, with implanted strips perpendicular each other, to read out indipendently two coordinates (x and y in our case, r{ and z in a typical collider experiment). The bulk of the detector is made by high resistivity n type silicon; the implanted strips are of p + -type on the junction side, and of n + -type on the ohmic side. The main problem related to this kind of technology is the presence of an electron accumulation layer at the interface between silicon and silicon dioxide, which dramatically lower the interstrip resistance on the ohmic side and practically avoid the measurement of the impact point of the ionizing particle crossing the detector. Essentially 2 dierent solution have been found to overcome this problem; the most widely adopted consists in the implantation of a blocking p + strip between n + strips on the ohmic side [13] (see Fig. 10.7), and the other one uses eld plates over the oxide to create the proper electric eld to increase the interstrip resistance [14]. This technology is now very well established, and a wide variety of complex detectors based on double sided silicon sensors is now in operation showing extremely good performances [15, 2, 16]. 107

110 counts efficiency: 90% E K [GeV] counts efficiency: 60% E K [GeV] Figure 10.5: Antiproton rates based on expectations of two models and the spillover background are shown for two 2 track reconstruction selection cuts. Exposition time of 1 year and Geometrical Factor of 25 cm 2 sr were assumed. 108

111 Anti-He/He up to E Kinetic energy per nucleon [ GeV/nuc ] Figure 10.6: The antihelium/helium integral ux measurement with the spillover background as a function of the kinetic energy per nucleon AC coupled detectors Another important progress that has been made in the last few years is the development of sensors with integrated on them the decoupling capacitors. Usually the signals coming from silicon detectors are capacitively decoupled to avoid the front end electronics integrating the leakage current of the sensors, thus worsening the signal to noise ratio and giving problems of saturation in the integrating stage of the electronics. Up to few years ago, the decoupling capacitors were realized on separate chips, which were wire bonded both to the sensors and to the front end electronics. This resulted in a more complex assembly phase and in a more critical environment, especially looking to spatial applications. The big R&D eort done in the last years made possible to integrate directly on the silicon sensors the decoupling capacitors, by separating the implanted strips from the readout metal strips by means of an insulating layer of silicon dioxide, with a typical thickness of 100/200 nm. In this way it is possible to obtain a suitable decoupling capacitance of the order of 20 pf/cm [17]. Two are the critical points of this process: the breakdown voltage of the capacitors and the percentage of defects (mainly short circuits between the implanted strips and the metal readout lines). The technology actually available provide devices with breakdown voltage greater than 100 V and a percentage of defects less than 1%, which are very well suited for our purposes. Moreover, in the proposed readout scheme for the PAMELA tracker, we will make use of optical decoupling to read out the signals coming from the ohmic side of the detectors, as will be explained in detail in a following section, in such away that the voltage needed to deplete the silicon will not be applied through the decoupling capacitor; this has both the advantage that the capacitors are not electrically stressed with a high voltage applied across them and that, in case of a short circuit across one capacitor, there is not a dramatic change in the electric eld in the region of the defect Double metal detectors Double sided detectors normally require to have the front end electronics mounted on two edges of them. This has many practical disadvantages in the construction of complex systems, related to the mechanical assembly of the detectors and to the amount of material which should be placed in front of the sensors, greatly enhancing the eect of multiple scattering. To overcome this problem, the L3 SMD collaboration, starting from 1991, developed a special fanout by means of a thin layer (50 m) of Kapton, with L{shaped metallized strips on it, which can be used to make the readout lines parallel on both sides [18]. Meantime, a novel technique based on a second metal layer deposition has been developed to solve this 109

112 Figure 10.7: Perspective view of a double sided silicon detector with the blocking p + strips on the ohmic side (from reference [13]). problem [19]. The rst metal layer is insulated from the second metal layer by means of a very thick oxide deposition (up to 5m); an ohmic contact between the two layers is then realized by means of an etching procedure. In this way, it is possible to run the top metallized lines at any angle with respect to the bottom ones, with a proper pattern of connections between the two layers. For example, it is also possible to connect two dierent strips in the bottom layer to the same strip in the top layer, allowing to realize in a simple way a complex geometric pattern of connections, which can for example help in reducing the total number of electronics front end channels [16]. This technique is now well established, and the silicon factories are able to provide double metal detectors with less than 1% of defects (a defect can be both a missed and a unwanted contact between the 2 layers); it is very well suited for the PAMELA tracker, because with this technique we can reduce the amount of material in front of the sensors and in the same time we avoid the mechanical problems related to the use of a Kapton fanout, which require special glueing technique in the assembly phase Double Sided Double Metal AC coupled technology for the PAMELA sensors Having in mind the situation sketched in the previous paragraphs, we choose to use double sided double metal AC coupled sensors for the PAMELA tracker, with p + blocking strips on the ohmic side. In this way we take the greatest advantage of the available technology, thus minimizing the problems that could arise from spatial environment. We avoid the use of external decoupling capacitors and Kapton fanout, reducing the problems of vibrations and stress in the launch phase. Great care has been payed in the design of the sensors that will be used to build the 6 planes needed for the tracker. First of all, a detailed analysis has been performed to optimize the geometrical dimensions of the sensors, having in mind the constraints related to the physical dimensions of the wafer used to build the silicon sensors. Most of the factories make use of \4 inches" wafers (10 cm diameter), even if the \6 inches" technology is in development and could eventually be used. To minimize the cost of the detector, we choose to use 6 sensors to build one plane, in such a way that the geometrical dimension of each sensor maximize the area used inside the 4 inches wafers; in this way the geometrical dimensions of each sensor result to be mm 2. The p + strips will be used to measure the x coordinate (on the bending plane), to have the better spatial resolution; the implantation pitch will be 25 m, and the readout pitch 50 m, to take advantage of the capacitive coupling between adjacent strips. On the ohmic side the implant pitch will be 67 m, and the readout pitch (using the double metal layer) 50 m, with the metal strips parallel to the junction's ones. Two sensors will be daisy chained, in such away that 3 ladders (each made with 2 sensors) will be used to build one plane. The total number of readout lines in one ladder will be 2048 (1024 for the x side and 1024 for the y side); for the whole tracker we will have = channels. Fig shows a schematic drawing of both sides of a Pamela ladder. The biasing mechanism that will be adopted is partly dictated by the request of the signal to noise ratio. On the bending side, we want to have the best possible spatial resolution, to measure as precisely as possible the momentum of the particles; the signal to noise ratio should be as high as possible. The biasing resistance should then be very high, to minimize the parallel noise coming from it. By adopting a biasing resistance of 50 M, with a characteristic shaping time of 1 s, the noise contribution coming from the biasing resistance is below 200 electrons [20], which is acceptably low compared to the MIP signal (25000 electrons) and to the main source of noise due to the strip capacitance (which is of the order of 300 electrons for a strip capacitance 110

113 of 20 pf using the front end chip VA1 that will be described in the following). X Side (Junction Side) mm Y Side (Ohmic Side) mm 67 µm pitch mm 25 µ m implantation pitch 50 µm readout pitch p+ implanted strips mm n+ implanted strips with p+ blocking strips 50 µm pitch readout strips mm mm 8 VA1 chips 8 VA1 chips mm Hybrid mm Hybrid 8 x 128 = 1024 channels 8 x 128 = 1024 channels Figure 10.8: Schematic drawing of both sides of a Pamela ladder. The biasing mechanism that can be used to provide such a high resistance are essentially the punch through [21] and the foxfet [22]. On the ohmic side, instead, the requests are not so tight, because the required spatial resolution is less stringent, and because the intestrip capacitance is intrinsically higher due to the double metal layer (' 40 pf for two daisy chained sensors). We can accept a resistance greater than 10 M, that can easily be provided also with a polysilicon deposition. Table 10.2 report the main characteristics of the sensors that will be used for the PAMELA tracker. Three factories have been contacted for the production of this kind of sensors: CSEM [3], Micron Semiconductors [11] and Hamamatsu Photonics [10]. We are actually investigating the reliability, the delivery time, the quality and prices of these sensors before taking the nal decision Spatial resolution and results of the tests With these kind of sensors and using the VA1 chips as front end electronics we could expect a signal to noise ratio considerably better than the one obtained with the L3 SMD ladders (which were read out by means of SVX chips, which have a sophisticated logic part to do on{line the zero suppression, but consequently have greater noise). Hence we can expect to reach a spatial resolution better than 7 m on the junction side and better than 15 m on the ohmic side, which permit us to fulll the physical goals above described. To study the achievable spatial resolution with these sensors and using the VA1 chips, we plan to do a test beam with the above described sensors at the PSI (Zurich) in october 1996, using both MIP and non MIP particles; we will have the possibility to use electrons, muons and pions with energies in the range 100{600 MeV. As a preliminary study, during summer 1995 we exposed a double sided, double metal AC coupled sensor into a beam test at CERN. The sensor was mm 2, and was produced by CSEM for the CMS collaboration [23]. It was equiped with 2 VA1 on both side, and read out by means of a custom designed electronic readout chain based on an optical decoupling system [24]. The results obtained were very satisfactory, showing a signal to noise 111

114 Main Parameters of Silicon Sensors for the PAMELA Tracker Double Sided, Double Metal, AC Coupled Geometrical Dimension mm 2 Thickness 300 m Leakage current < 3A Decoupling capacitance > 20 pf/cm Breakdown voltage of capacitors > 20 V Total number of defects < 2% p side Implant strip pitch 25 m Readout strip pitch 50 m Number of readout strips 1024 Biasing resistance > 50 M Interstrip capacitance < 10 pf n side Implant strip pitch 67 m Readout strip pitch 50 m Number of readout strips 1024 Biasing resistance > 10 M Interstrip capacitance < 20 pf Table 10.2: Main parameters of the silicon sensors for the PAMELA tracker. ratio of 27 on the junction side and 20 on the ohmic side. The distributions relative to the charge integrated on both sides of the detector are shown in Fig. 10.9, while Fig shows the correlation between the charges collected on the two sides. Even if the geometrical dimensions of the tested sensors are not the same as the nal ones, this test is a good indication that we can achieve the expected spatial resolution Mechanical Design and Construction of PAMELA Tracker This section will present the mechanical structure that will be used to build up the Pamela tracker, together with the simulations that have been carried out to investigate the deformations of the structure during the launch phase. Particular care has been payed in the design to reduce the total radiation lenght of each plane, to improve the physics performances of the detector Description of the Mechanical Support Design The mechanical support of each silicon plane has been designed having in mind the stringent requirements needed during the launch phase; moreover, an important feature is the modularity of the system, because we want tohave inany time before the launch the possibility tochange one silicon plane in case of any kind of problems. A silicon plane is composed by 3 ladders, each made by 2 silicon sensors 7 cm long and by an aluminium oxide hybrid 5 cm long, containing the front end electronics. Each ladder is stiened by means of 2 carbon ber bars, precisely glued on the lateral sides of the ladder. These bars are 300 m thick and 5 mm height, and run all along the ladder, for a total lenght of 19 cm. The three ladders of each plane are glued together, and inserted inside a mobile carbon ber frame which provide the necessary rigidity of all the plane. A renforcing additional stiner is positioned over the hybrids. Fig shows a perspective view of this mechanical arrangement, whose total thickness is 8 mm; this structure is then inserted into a xed aluminium frame that is directly connected to the 4 columns that are supporting the entire magnet structure, which is shown in Fig Once the mobile frame is completely inserted into the xed one, it is kept pressed against two reference sides of the xed frame through a pin&spring system. The tolerances and any possible displacement occurring due to the vibrations, including the displacements on z axis, are therefore recovered by this system Calculations on the Structure The calculations on the above decribed structure were carried out using the FEM system with the spectral load expected during the launch phase, which is described in Appendix A. Fig shows the subdivision of one 112

115 Junction Side S/N= Integrated Charge (Arbitrary Units) Integrated Charge (Arbitrary Units) Ohmic Side S/N= Figure 10.9: The Landau distributions corresponding to the charge integrated on both sides of the tested detector. Charge on Y Side (Arbitrary Units) X-Y Correlation Charge on X Side (Arbitrary Units) Figure 10.10: Correlation between the charge integrated on the junction and ohmic side of the tested detector. 113

116 Figure 10.11: Perspective view of the carbon ber structure supporting the 3 ladders of each tracker's plane. Figure 10.12: Perspective view of the tracker system, composed by the 5 sections of the magnet and the 6 silicon planes. 114

117 half of a silicon plane into the unitary elements; the numbering of the nodes is also shown. The calculatiions were carried out only for one half of a plane because of the symmetry properties of the system. Figure 10.13: Subdivision of one half of a single tracker plane frame into the unitary elements. The numbering of the nodes is also shown. The results of the calculations on this structure show that the rst peculiar vibration frequency of the system (resonance frequency) is 445 Hz, suciently high to avoid major problems during the launch phase. Acceleration Analysis The eect of a static acceleration of 8g along the 3 X, Y, Z axes has been studied with the same FEM program. During the acceleration along each axis, the displacements of the three components (X,Y,Z) of each node have also been studied. The major displacement has been observed for the acceleration along the Z axis: it was about 16 m in the Z direction at the center of the plane (Node 490). The corresponding maximum stress was computed to be 0.05 N/mm 2. The results of this simulation are reported in Fig Shock Analysis The eects of a shock with 40g load for 1 ms have also been simulated in the same FEM framework. The calculation model made use of a fake displacement of the carbon ber frame. The maximum relative displacement of 184m has occurred at the Node 966, while the displacement of the center of the plane (Node 490) was below 100 m (Fig ). The maximum stress of the center of the plane was calculated to be 0.3 da N/mm 2. The same calculations were carried out for a shock with 50 g load for 1 ms; the maximum displacement resulted to be 283 m for Node 966 and below 100 m for the center of the plane, with corresponding maximum stress 0.43 da N/mm 2 (see Fig ). Vibration Analysis The calculations have been carried out using a resonance vibration frequency of 445 Hz with 1% internal quenching. The diagram given in Fig show that the major displacements (' 5m) are occurring at the external nodes of the frame (Node 966) Alignment of the detector In order to achieve the expected momentum resolution of the PAMELA tracker, we need to know very precisely the relative position of all the silicon sensors (alignment); in particular, our goal is to know the position of all the detectors with a precision better than 1 m, in such away that the silicon spatial resolution is not spoiled by the misalignment eects. To investigate this problem, we developed a simplied alignment algorithm based on a 2 method, that allows for all the 6 detector planes the reconstruction of the relative x and y displacements 115

118 Figure 10.14: Displacement of the z component of the various nodes of the silicon plane under a static 8 g acceleration. (dx, dy) and of the rotation angle in the x{y plane ( xy ). By means of a certain number of tracks crossing our detector, we determine for each plane the displacement parameters by minimizing the squared sum of the distances between the expected crossing point on the plane and the measured ones. We use for this both the x{z and y{z projections. We applied this algorithm to simulated events generated with a modied geometry with the planes displaced with respect to their theoretical positions. We found that, assuming the relative position of the 6 silicon sensors in one plane is known from the assembly phase, approximately 1000 events are enough to reach the expected alignment precision, and we didn't observe any systematic eect on the obtained parameters. We can thus assume that in about 100 minutes of run at the beginning of the data taking we will align our detector to the expected precision, by using high energy protons (E>20 GeV), which are less sensitive to the scattering eects Readout electronics This section will describe the readout electronics which has been designed for the PAMELA tracker, taking a particular care in the minimization of the power consumption, which is one of the most important constraint present in the project. First, the front end and the hybrid will be described, followed by the analog to digital conversion part with the optical decoupling system. Then, we will present the processors used to compress the data, the interface with the main computer of the experiment and the sequencer used to generate the signals needed to handle the detector. Fig , and show the schematic drawings of all the relevant parts of the readout electronics, that will be explained in detail in the following Front end A major issue in the development of silicon micro-vertex detectors, is the spatial resolution. Since the spatial resolution of a detector is strongly correlated to its signal to noise (S/N) performance, over the last few years a big eort has been put into the development of low noise front-end electronics. As a result of these eorts, recently a new generation VLSI chip called \VIKING" [25, 26] has been designed and constructed and at present it is successfully used in many applications. The VIKING chip consists of 128 charge sensitive preampliers and 128 CR-RC shapers (typically with 1 s shaping time) followed by a sample and hold circuitry, input and output multiplexing and a dierential output buer, with less than 10 ns settling time to drive the analog information. A schematic drawing of the circuit is shown in Fig The bonding pitch on the amplier input is 47 m, allowing a direct bonding to the detectors with readout pitch of 50 m. The use of time continuous ltering enables asynchronous triggered applications and very good signal to noise ratios. It has been possible to reach with small good detectors and the VIKING chip a signal-to-noise ratio of 77 and a spatial resolution of 1.2 m [27, 28]. 116

119 Figure 10.15: Shock analysis for 40g load of 1 ms duration. The displacements and the stress are shown for various nodes. 117

120 Figure 10.16: Shock analysis for 50g load of 1 ms duration. The displacements and the stress are shown for various nodes. 118

121 Figure 10.17: Vibration analysis for resonance frequency of 445 Hz. The gure shows the displacements of the nodes on the main symmetry axis of the silicon planes. With new VLSI design techniques developed for the LHC at CERN an improved version, based on the same principle as the Viking, has been designed and produced; this version is called VA1, and is actually available commercially from IDEAS [29]. For the full 128 channel version of VA1 a noise performance of ENC =165+6:1e, C (10.1) has been measured, using a shaping time of 2 s, where ENC is the equivalent noise charge in electrons and C is the capacitive load of one strip of the detector in pf. Using 1 s shaping time, the noise performance become slightly worst: ENC = :5e, C: (10.2) The chip works with +/-2 V supply voltage, and the output multiplexer can work up to a frequency of 5 MHz, even if we plan to use a readout frequency of 1 MHz to reduce the power consumption. Fig shows the comparison of the Viking and VA1 noise as a function of loading capacitance for 2 s shaping time. With the silicon sensors which were described above, using 1 s shaping time we will expect to have an ENC of 400 e, on the junction side and 550 e, on the ohmic side, including the contributions of the biasing resistance and of the load capacitance, hence giving a possible signal to noise ratio ' 60 and 40 respectively on the two sides, provided that the other parts of the readout electronics do not introduce additional noise. The dynamic range of the VA1 has been measured to be +/- 10 MIPs; this feature is very important, because it allows to separate light nuclei performing de/dx measurement in the 6 detector's planes. With this dynamic range it is possible to identify light nuclei up to Berillium (Z=4), because the energy released by ionization is proportional to Z 2, without suering of saturation eect. The last important point concerning the VA1 is the power consumption, which is a very critical part in any spatial project. The analog part of the VA1 (in particular the input FET) is responsible of the greatest power consumption, since it should be continously biased. The design value of the drain current is 500 A, which gives over 2 Volts drop a power consumption of about 1 mw for each preamplier. Including the other parts of the chip, the total consumption reaches 1.2 mw per readout channel. Anyway, we did some measurement reducing the drain current and measuring the ENC and the signal to noise ratio; the results of these tests are reported in Fig , with 2 s shaping time. We observe that reducing by a factor 2 the drain current, (hence reducing almost to a half the power consumption of single front{end channel) the S/N ratio worsen by only 5%. We can in this way operate the VA1 with a power consumption of 0.7 mw/channel, without any important degradation in the performances of the tracker. 119

122 GENERAL STRUCTURE OF PAMELA TRACKER 6 double sided silicon sensors per plane = 36 sensors 8 VA1 per each side of a ladder = 8 x 2 x 3 x 6 VA1 = 288 VA1 = channels Hybrid (double sided) Ladder (2 sensors = 1 ladder) 1 VA1 channel = 0.7 mw Total power for the VA1: 0.7 x mw = 26 W Figure 10.18: Schematic view of all the PAMELA tracker, with the expected power consumption for the VA1 front end chips. 120

123 HYBRID 1 Double Sided Hybrid / ladder ADC Section 2 ADC Section / ladder (1 for X, 1 for Y) 5 cm cable ADC Serial Out 8 VA1 30 mw 15 mw Shift_in (8) Hold (1) Clkv (1) ADC Control Lines 40 mw Control Logic 15 mw HCPL 40 mw x 2 = 80 mw Control Clock Readout time / channel = 1 µsec HCPL = Digital Optocouplers Total readout time = 1 msec (1024 channels) Power needed for 1 Hybrid + 2 ADC Section = 360 mw 360 mw x 18 = 6.5 W Figure 10.19: Schematic block of a possible conguration for the hybrid and the ADC section. DSP Section for Data Compression ADC Section ADC Section Serial Link Serial In Parallel Out CMOS Logic 30 mw ADSP mw 33 MIPS 80 KByte Serial Link to DAQ ADC Section 1 DSP Section / 3 ADC Section = 2 DSP Section / Plane = 12 DSP Section 330 mw x 2 = 4 W Figure 10.20: Schematic block of the readout board containing the Digital Signal Processor to do the on{line data compression. 121

124 Test_on Ckb Shift_in_b Pad Preamp Shaper S&H Pad Preamp Shaper S&H 128 cell bit-register 128 x Analogue Mux Pad Preamp Shaper S&H 128 x Analogue Mux 128 cell bit-register Pad Preamp Shaper S&H Pad Preamp Shaper S&H Shift_out_b Calibration Analogue bias & adjustment Hold Differential Analogue Output Figure 10.21: Block scheme of tha VA1 and Viking chip. ENC (rms e-) ENC vs Load Capacitance for 2 microsec shaping VIKING ENC(rms e-) = 135 e e- / pf VA1 ENC(rms e-) = 165 e e- / pf C_load Figure 10.22: Comparison of the Viking and VA1 noise as a function of loading capacitance. 122

125 PAMELA ladder with VA1 frontend S/N I_drain (ma) Figure 10.23: Bending plane side signal{to{noise performance of AC-coupled (to VA1) PAMELA ladder as a function of chip drain current (power consumption) Design of the hybrid The hybrids which will be used for the PAMELA trackers are double sided devices containing 8 VA1 on each side. In this way, each hybrid will serially manage 1024 channels on both sides; each plane will be served by 3 hybrids. Aschematic view of the hybrid is shown in Fig ; all the relevant parts will be presented in the following. The main purpose of the hybrid is the housing of the front end chips. It will be made by a core of 300 m thick Aluminum Oxide, which has the necessary mechanical properties to ensure a good stability ofthe bonding wires and a good thermal dissipation. The two opposite sides of the aluminum oxide will be processed with a special procedure to deposit on them all the necessary strip to interconnect the VA1 with the readout electronics, with a pitch down to 200 m to match the characteristics of the VA1. Normally the VA1 can be daisy chained directly on the hybrid by wire bonding the logic signals used to handle the output shift register of the chip. In particular, the readout of the chip is initiated by an input signal (called shift in), and the chips send out an output signal (called shift out) when all the 128 channels have been read out. The usual procedure to daisy chain various chips is to wire bond the shift out of one chip with the shift in of the following one; in this case it is necessary to send from the external control logic only the shift in of the rst chip. However, this procedure can be very dangerous in case one chip breaks, without sending out the shift out signal. The chain in this case will be interrupted, and up to 8 VA1 chips could be lost in our case by adopting this solution. The simpler solution that we found to overcome this problem is to reduce the numberofva1 chips daisy chained, sending more control signals form the external logic. For example, we could daisy chain only 2 chips, sending from the external 4 dierent shift in signals, or we can send 8 dierent shift in signals from the external, without doing any daisy chaining; in this way, in case of any problem with one chip, we will loose at maximum 1or2VA1 chips. The disadvantage of this solution is that the connections between the hybrid and the external control logic become more complexes; this is the solution adopted in the schematic drawing of Fig Meantime we are studying a dierent solution that relies on the use of a programmable control logic to be placed on the hybrid itself. This control logic will decode an input address and consequently will provide all the needed signals to the relevant VA1 chips. On the market we can nd programmable logic device with spatial qualication; for example ACTEL and CYPRESS produce non volatile Field Programmable Gate Arrays (FPGA) with military qualication, containing up to ' 250 Input{Output pins, with extremely low 123

126 power consumption 1, which are well suited for our purposes. The disadvantage of this solution is that the hybrid itself become more complex. Another important function of the hybrid is to provide all the voltage references and current sources necessary for the VA1. This task is accomplished by a passive network of resistors (eventually laser trimmered to improve their precision) and ltering capacitors. Moreover, the hybrid will do the sum of the dierential analog signals coming from the 8 VA1, and will house a dierential line driver to send the analog signals to the ADC located on a separate board on the edge of the magnetic screen, ' 10 cm away. We selected few operational amplier that can be used for this purpose, with a good output current capability, lowpower consumption and proper gain{bandwidth product. Some of these components are listed in Table 10.3, together with their relevant properties. Device Power cons. Gain{Bandwidth Output current Noise (mw) (MHz) (ma) (nv/ p Hz) AD AD AD AD MAX Table 10.3: Main characteristics of some operational ampliers that could be used to drive the analog signals from the VA1 to the ADC. Each hybrid will be connected to the ADC board by means of a exible screened kapton cable, soldered on the hybrid itself because of the small available space inside the magnet gap, and with a proper micropitch connector on the other side; we plan to use the connectors from MOLEX [30] which are designed to be used together with screened kapton cables produced by PARLEX [31]. The connectors are provided with metal strain relief tabs to connect to the cable with a Zero Insertion Force system; the shielding can be applied to one or both sides of the cable. We plan to order a prototype hybrid from the italian factory MIPOT [32], that has a proper technology to realize this kind of devices. We are also investigating the possibility to use the hybrid directly produced by IDEAS [29] which will provide them assembled with the VA1s and completely tested Analog to Digital Conversion The analog to digital conversion of the analog signals coming from the VA1 is performed on an electronics board (called readout{board) housed immediately outside the magnetic screen, which contains also the Digital Signal Processors that will be described in the next paragraph. We plan to realize one board per each plane of sensors, which will house 6 ADC section (2 for each ladder, 1 for the x{side and 1 for y{side signals) and 2 Digital Signal Processor section (1 DSP every 3 ADCs is a reasonable estimate, as will be explained in the following). This schematic arrangement is shown in Fig We now describe the main feature of the ADC section, that will provide both the Analog to Digital Conversion of the VA1's analog signals and the generation of all the relevant digital signals for the VA1 and for the ADC by decoding a control clock coming from a dedicated sequencer board that will be described in a following section. The generation of the needed digital signals is realized by means of a low power Control Logic realized with the FPGA devices already described in the hybrid section (15 mw). Only 1 control clock is needed to provide all the relevant signals both for VA1 (8 dierent shift in, 1clock and 1 hold) and the ADC (a dierent number of lines, depending on the specic ADC that will be used). The dierential analog signal relative to8va1 (1 ladder, x or y projection) is received by the same type of operational amplier used on the hybrid, and sent to the analog input of the ADC, after some level adjustment as required by the ADC itself. The ADC should operate with a sampling rate of at least 1 MHz, to follow the signals coming from the hybrid, and it should be able to handle the dynamic range of the VA1 (20 MIPs). Since we could expect to have a signal to noise ratio of 50 for a MIP particle, we need at least a 10 bit ADC (1024), to avoid to introduce an additional noise due to the analog to digital conversion itself. Moreover, the power consumption should be kept as low as possible. In the last few years, thanks to the new low power CMOS techniques developed by the electronic components factories, ADC which are well suited for our purposes appeared on the market. After a careful analysis, we selected and tested the devices which are shown in Table 10.4 as possible ADCs for the PAMELA tracker. They 1 The power consumption of these CMOS devices critically depends on the frequency of operation. Foratypical operation frequency of 1 MHz, the power consumption can be kept below 10 mw/chip. 124

127 can work between -2 V and +2 V, hence greatly simplifying the power supply system, since we can use the same power lines used to bias the VA1 chips. Device # of bits Sampling rate Power consumption Output (MHz) (mw) (@1 MHz) MP87L Parallel CLC Parallel SPT Parallel MP87L Serial SPT Serial Table 10.4: Main characteristics of selected ADCs. All these devices meet the required specication concerning the dynamic range and the sampling frequency, and we can notice that the power consumption is extremely low, well below 100 mw for each of them. Another important feature is that few of these ADCs have a serial output for the data; this characteristic could be very useful for our purposes, since we plan to optically decouple the data coming from the detector both to minimize the interference induced by the other parts of PAMELA and of the satellite on the tracker, and to avoid to stress the decoupling capacitors integrated on the sensors with the voltage needed to deplete the silicon. Following the strategy used for the rst time for the L3 SMD, we will use as a reference level (`ground') for the ADCs the potential of the silicon sensor's side to which the ADC is connected, as schematically shown in Fig The digital data which are the output of the ADCs are then optically decoupled by means of Power lines for the Y side electronics Power lines for the DAQ electronics Bias for the Silicon ~ 50 V Ohmic Side Signals from Y side Silicon Amplifier + ADC DAQ System Junction Side Amplifier + ADC Signals from X side Power lines for the X side electronics Optical Decoupling of digital data Figure 10.24: Schematic view of the adopted electrical scheme to readout the pamela tracker, with the optical decoupling system to minimize the noise gure of the system. digital optocouplers to be trasferred to the data acquisition system, which has as reference a dierent potential. In this way, there is not a direct electrical connection between the front end and the data acquisition system, and the noise induction on the signals coming from the silicon is minimized. The use of a serial output ADC simplify the optical decoupling, avoiding the use of a big number of optocouplers or of a dedicated logic for the serialization of the signals. In the selection of the digital optocouplers, particular care has been payed in the reduction of the power consumption and in the achievable transmission rate. The most advanced producer in this eld is Hewlett Packard; a selection of the devices with lowest power consumption is shown in Table The greatest part of the power consumption of these devices is related to the digital output logic of the chip, that is not normally realized in CMOS technology. Thus, to further reduce the power consumption, we are investigating the possibility to realize a custom designed digital optocoupler starting from a pair of optically coupled LED and photodiode (which are commercially available) and a CMOS low power comparator on the output stage. We are actually doing some tests with devices from MAXIM Coorporation (MAX907, MAX908 and MAX909), with a typical power consumption less than 5 mw/comparator. The block scheme of the ADC board is reported in Fig ; all the logic signals are decoupled by means of 125

128 Device Power consumption Transmission rate (mw) (MBaud) HCPL HCPL HCPL HCPL HCPL HCPL Table 10.5: Main characteristics of the selected digital optocouplers. The power consumption slightly depends on the logical input level. low power consumption optocouplers. The serial output of the ADC is then sent to a Digital Signal Processor, to do a rst on{line compression of the data. By operating at 1 MHz readout frequency of the VA1, given the 1024 modularity, we need approximately 1 ms to read out all the tracker. This time is small enough to avoid introducing big dead time, since the average trigger rate of the experiment is expected to be 5 Hz Preprocessing of data with a DSP / Data Compression A peculiar characteristic of any experiment in the space is the limited bandwidth for the trasmission of data to earth. The foreseen telemetry system for the PAMELA detector will allow to transmit Mbit/sec when the comunication with the earth is open. Normally, each session will last approximately 10 minutes, and a variable number of session per day is foreseen, ranging from a minimum of 1 to a maximum of 4. In these condition, the PAMELA detector will be able to transmit from a minimum of 9 Gbit/day, to a maximum of 36 Gbit/day. The expected average rst level trigger rate (excluding the brasilian anomaly) is approximately 5 Hz, giving a total number of ' 1 Milion events/day. The amount of data collected by the tracker alone will then correspond to approximately 80 GByte/day (assuming 1 word of 16 bit for each of the channels), completely outside of the characteristic of the telemetry system. It is then clear that a big reduction or compression of data is needed on board of the satellite 2. The most ecient way to accomplish this task at the level of the tracker is the use of a series of dedicated CPUs, which receive the data directly from the ADCs and apply some compression alghoritm, which can be very ecient, since the event topology for the majority of the events is very simple (1 single track crossing 6 planes, giving a very small number of clusters identied in the detector). Since we know the shape of the signals coming from the VA1 (pedestals), the best way to do the data compression is by using the so called xed aperture Zero Order Predictor (Z.O.P.). This technique starts from the last transmitted value to predict the following value, and this value is transmitted only if the real data is out of the tolerance band around the predicted one given by the xed aperture. In our case the shape of the signal (after the pedestal subtraction and the elimination of the noisy strips) is at, except for the ionization clusters, so we nd the predicted value for the following strip adding to the last transmitted value the dierence between the two related pedestal values. When the real value is outside a xed band from the predicted one, it is transmitted together with its address and the procedure restarts. We tested this method with the data gathered in the test run of summer 1995 using the double sided double metal detector described in section We tested the alghoritm using dierent values for the xed aperture; the results are given in Tab These preliminary results show that the methods can achieve a good compression factor; it can be improved using its good cluster identication capability to reduce the distortions by a full transmission of the cluster. The quantity of needed CPUs to accomplish this data compression depends on their characteristic, which are a compromise between computational speed and power consumption. The best compromise that we found is represented by the Digital Signal Processor ADSP2181, from Analog Device. This is a very advanced device, integrating on board 80 kbyte of memory, which works with a 33 nsec period clock, hence reaching 33 MIPS performances. It has 2 serial input{output ports, 1 special IDMA ports for high speed asyncronous access to the internal memory, a byte DMA port and can manage 6 external interrupts. The power consumption during normal running amounts to 300 mw and can be reduced down to 30 mw in idle mode, waiting for an interrupt to restore normal working conditions. The internal memory is congured as 16 KWords of Program memory (24 bits) and 16 KWords of Data memory (16 bits). With such an high memory buer, it is possible to avoid the use of any temporary memory buer (like FIFO, for example), with a great reduction in the power consumption. 2 The reduction of data using a second level trigger is discussed in Section

129 Results of data compression. 3 cut 2 cut 1 cut 0 cut y side compression 96.2% 91.7% 76.2% 27.4% x side compression 96.5% 91.1% 74.1% 22.9% Variation in cluster energy (%) 2.98% 0.71% 0 0 Variation in max. strip position (%) Variation in cluster width (%) 1.34% 0.42% 0 0 Table 10.6: Results of the compression procedure: rst the obtained compressions for both x and y side for dierent values of the xed aperture are given; then the distorsions of some relevant cluster quantitities for the same values of the cuts are reported. The event will be processed \on{line" with the described compression alghoritm, and will be sent to the main CPU for any further analysis (second level trigger) few microseconds after the end of the readout. The computational power and the internal memory of this DSP allows to use 1 device every 3 ADCs blocks, corresponding to channels; in this way, we will use a total of 12 ADSP2181 for all the tracker. In the case we will use the ADCs with serial output, we will need to build a dedicated logic to do the serial-to-parallel conversion of the data, since the serial port of the DSP is not able to receive 3 12 = 36 bits in 1 s (the maximum allowed speed is 50 nsec/bit). The conversion logic will be built using the ACTEL or CYPRESS programmable logic device already described in the hybrid section; in this way, we will introduce an additional power consumption of 30 mw for each DSP. The input of the data in the DSP memory is done in a completely asyncronous way by means of the special IDMA port of the device; through this port we can access to the internal DSP memory without interfering with the normal operations of the device. One of the two serial port of the DSP will be used as a bidirectional link with the main CPU, to download programs inside the DSP and to send the compressed data to the main DAQ system. In this way, we will need 12 separate input/output units on the main DAQ system to communicate with the tracker's processors Sequencer All the logic signals that are used to control the tracker's electronics are generated on the ADC section with the already described dedicated FPGA device; this device decode a complex control clock, that is generated by a sequencer unit, one for all the tracker system. This unit essentially receive the trigger signal (or the calibration signal during some particular runs) from the trigger system and generate a complex pattern of signals, that will be passively decoded on the ADC sections. The signals that are needed for the whole system are schematically reported in Fig The rst signal that should be generated is the hold signal, used to store inside the VA1's capacitors the voltage level corresponding to the charge integrated by the various strips. This signal should have a precise timing with respect to the crossing time of the particle that generated the trigger, to sample the signal in it's maximum. The hold signal should be sent also to the DSPs to inform them of the starting of a readout chain. After the charge have been stored, the output multiplex should be clocked, to serially read out the analog signals corresponding to all the channels. As already explained, the VA1s need a shift in signal to address a particular chip and a clock signal to multiplex the analog out. The Control Logic of the ADC sections should generate dierent shift in signals for the dierent VA1s, according to the choosen VA1 modularity. The conversion signal for the ADCs (clock ADC) is generated with a xed delay with respect to the clock signal, to sample the analog signal when it is stable (typically after few hundred nanoseconds). The clock signal, used to multiplex the analog output with 1 MHz frequency, has also the function of triggering the logic used to do the parallel-to-serial conversion, and to inform the DSPs to sample the input data through the IDMA port. All these signals are generated in a syncronous way, using one main clock, to avoid glitches and unwanted delay between them. The sequencer send to the main trigger board a ready signal when the readout of all the 1024 channels has been nished (approximately 1 msec after the trigger). The sequencer is located in the electronic box that houses all the trigger logic; it is based on an FPGA device similar to the one already described for the hybrid, that guarantee high performances with extremely low power consumption (30 mw). It will have dedicated fan{out units, to send signals to all the 36 hybrids, 36 ADCs and 12 DSPs present in the system. Each fan{out is realized with low power CMOS dierential line drivers, with a typical power consumption of 20 mw/channel. 127

130 End of the readout TRIGGER 1 msec after the trigger Hold T=2 µ sec after the trigger Shift_in 1 Shift_in 2 Shift_in Clock for the VA1 T=1 µsec Clock for the ADC Figure 10.25: The signals needed to control the tracker's readout electronics, generated on the various ADC sections by decoding the Control Clock of the sequencer board. 128

131 Summary on power consumption As a summary, Table 10.7 report schematically the modularity and the expected power consumption (slightly depending on the choosen components) of all the electronics related to the tracker. Power needed for the redundancy of the system is not included. Summary of the tracker's electronics Front{End 8326 VA1 = channels 0.7 mw W Hybrid (36) Line Drivers 30 mw 36 1W ADC (36) Line Receivers 15 mw W Control Logic 15 mw W ADC 40 mw W Optocouplers 80 mw W DSP (12) Serial-to-Parallel Logic 30 mw W DSP 300 mw W Sequencer (1) Sequencer Logic 30 mw Fan{Out 20 mw W Total Power 37 W Table 10.7: Summary of the power consumption for the whole PAMELA tracker Mechanical positioning and geometrical dimensions of the electronic boards As has been described, the electronics for the whole tracker consists essentially in the hybrid with the front-end chips, in a complex printed circuit board containing ADCs and DSPs (called readout{board), and in a separate sequencer board. The hybrids are directly tied to the sensors, and are located inside the mechanical structure supporting the detector's planes, already described in the mechanic's section. This structure is inserted inside the gap between two adjacent pieces of the magnet. The hybrids are 300 m thick and 53.3 mm wide, to match the silicon geometry; their lenght is 50 mm. The screened exible kapton cable connecting the hybrid to the readout board is soldered on the hybrid side, to avoid the insertion of a connector in the small available space inside the gaps of the magnet. The readout boards (1 for each sensors plane) are located immediately outside the magnetic screen and the anticoincidence counters, approximately 5 cm away from the hybrid. These boards will be built using the avionic technology, in a multilayer structure with thermal dissipating planes inside, and will be ' 2 mm thick. The thermal contact with the magnetic screen ensure the dissipation of the heat produced by ' 1.5 W for each board. The board will be housed in the available space between two adjacent sections of the magnet, rotated by 90 with respect to the hybrids (' 16 9cm 2 ). All the 6 boards (the 6 board will be placed on the top of the 5, since the sections of the magnet are 5) will be included in the same metallic box, provided on one side by the screen of the magnet, and on the other side by a suitable cover, realized with 1 mm thick aluminium. This box will ensure a proper electromagnetic screen for the electronics, reducing both interferences from the outside world on the tracker and vice versa. The readout boards will be connected with the other parts of the satellite only with the serial links for the data transmission and for the slow control system, with the cables from the sequencer and with the power supplies cables. 129

132 130

133 Bibliography [1] R.Battiston et al., The SMD Study Group CERN{LEPC 91-5, LEPC4{Add. 1, April 199 [2] M.Acciari et al., Nucl. Inst. and Meth. A351 (1994) 300. [3] CSEM, Centre Suisse d'electronique et de Microtechnique S.A., Maladiere 71, Case Postale 41, CH-2007 Neuch^atel, Switzerland. [4] A.Adam et al., Nucl. Inst. and Meth. A344 (1994) 521. [5] O.Adriani et al., Nucle. Phys. B (Proc. Suppl.) 32 (1993) 480. [6] O.Adriani et al., Nucl. Inst. and Meth. A342 (1994) 181. [7] Review of Particle Properties, Phys. Rev. D50 (1994) [8] P.Kiraly et al.,1981, Nature, 293, 120. [9] F.W.Stecker, A.J.Tylka, Ap. J., L51(1989)336. [10] Hamamatsu Photonics, Shimokanzo, Toyooka-village, Iwata-gun, Shizuoka-ken, Japan. [11] Micron Semiconductor Ltd., 1 Royal Buildings, Marlborough Road, Churchill Industrial Estate, Lancing, Sussesx BN15 8UN, England. [12] SINTEF, Postboks 124 Blimdem, N-0314 Oslo, Norway. [13] G. Batignani et al., Nucl. Instr. and Meth. A257 (1987) 587. G. Batignani et al., Nucl. Instr. and Meth. A277 (1989) 147. G. Batignani et al., Nucl. Instr. and Meth. A310 (1991) 160. [14] B.S. Avset et al., IEEE Trans. Nucl. Sci. Vol. 36, No. 1 (1989). [15] B. Mours et al., \The design, construction and performance of the Aleph silicon vertex detector", CERN- PPE/ [16] V. Chabaud et al., Nucl. Inst. and Meth. A368(1996)314. [17] H. Dijkstr et al., IEEE Trans. Nucl. Sci. Vol. 36 (1989) 591. I. Hietanen et al., Nucl. Inst. and Meth. A310 (1991) 671. [18] D. DiBitonto et al., Nucl. Inst. and Meth. A338 (1994) 404. [19] D. Pitzl et al., Nucl. Inst. and Meth. A348 (1994) 454. K. Saito et al., Nucl. Inst. and Meth. A326 (1993) 204. [20] E. Gatti and P.F. Manfredi, La rivista del Nuovo Cimento, Vol. 9, Serie 3, Numero 1 (1986). V. Radeka, Ann. Rev. Nucl. Part. Sci., 38 (1988) 217. [21] J. Ellison et al., IEEE Trans. Nucl. Sci. NS-36 (1989) 267. [22] P.P. Allport et al., Nucl. Inst. and Meth. A310 (1991) 155. M. Laakso et al., Nucl. Inst. and Meth. A326 (1993) 214. [23] CMS Collaboration, Technical Proposal, CERN/LHCC [24] O. Adriani et al., \A data acquisition system for silicon microstrip detectors", Dipartimento di Fisica and INFN Firenze Preprint, DFF /

134 [25] F.Nygard, et al., Nucl. Inst. and Meth. A301 (1991) 506. [26] O.Toker et al., Nucl. Inst. and Meth. A340 (1994) 572. [27] J.Straver et al., Nucl. Inst. and Meth. Phys. Res. A348 (1994) 485. [28] P.Weilhammer, et al., Nucl. Inst. and Meth. A342 (1994) 1. [29] IDE AS, Gaustadalleen 21, N-0371 Oslo, Norway. [30] Molex Incorporated, 2222 Wellington Court, Lisle, Illinois, 60532, USA. [31] Parlex Corporation, 7 Industrial Way, Salem, New Hampshire, 03079, USA. [32] MIPOT S.p.A., Via Corona 5, Cormons (Go). 132

135 Chapter 11 Calorimeter 11.1 The physics task The main task of the Pamela calorimeter, having an high transversal and longitudinal granularity, is to give a substantial contribution to the identication of the antimatter signal. This means to identify the positrons in the physics background generated by the proton ux and, at the same time, to extract the antiproton signal in presence of a comparatively high electron ux. The tools that the calorimeter makes available are bound to the total energy deposition in the calorimeter and the transversal and longitudinal shape of the measured shower development. For the detection of positrons we must take in to account the fraction of the converting hadrons whose shower shape pass the selection criteria for true electromagnetic showers, i.e. originated in the rst calorimeter layers and collinear with the impinging track. The expected performances of the Pamela calorimeter are: allow to extract the antiproton signal in presence of the high electron ux, (90% eciency, rejection 10,3 {10,4 ) identify positrons in the physics background generated by the proton ux above 1 GeV, rejection of 10,4, eciency 70% The starting point Since 1990 a project was started within the INFN WiZard collaboration for the development of Si{W calorimeters to be applied in the space station on balloons and satellites. The achieved milestones are the following: 1991{1992 test beam of the rst calorimeter and trimming of the MonteCarlo code on the results; 1993 NSBF balloon ight TS93 from Fort Sumner, e data published; 1994 NSBF balloon ight Caprice, from Lake Lynn, antiprotons preliminary data available; we plan to use the well working instrument for more balloons ight; 1995 the NINA project starts, today the Q-model is integrated and indicates the ability to perform the selected physics tasks (Launch rst half 1997); 1996 Start of the detailed work of detector simulation and prototypes production of the Pamela spectrometer whose key task is to give decisive insights about e +, antiprotons and antimatter in cosmic rays. Coming to the calorimeter, the original Si detectors and electronics are constantly optimised. Initially large batch of detectors (100) were purchased by all available vendors to evaluate the ability to deliver on time satisfactory material. The detectors have been carefully studied and kept under control at all phases of the project. Within the two balloon ight TS93 and CAPRICE, more than 5 m 2 of silicon detectors have been put to work and, after recovery, are ready to y in the coming balloon experiments. During the construction of the NINA telescope the batch of very high quality of detectors, especially chosen, and the detector's mounting modules have been qualied for the use in space following ESA{NASA standards. 133

136 While developing the NINA telescope we have worked in tight contact with the industry and collected a fair experience for the production of hardware to be used in space. The space qualication of NINA has allowed us to face the various systems and organisational problems typical of space applications. Figure 11.1: the experimental distribution of a well discriminating variable related to the starting point of the shower for a) electrons and b) protons. This variable correspond to the path length of the particle up to the interaction point weighted with the distance of this starting point from the top of the calorimeter Balloon ight results Here we report the baseline structural details of the two ying instruments TS93 and CAPRICE 94, together with the characteristics measured during the ight. The calorimeter is used as a tracking{showering device. For the TS93 ight we prepared 5 x{y Si layers interleaved with tungsten for a total of 4 X 0. The reduced thickness, in terms of X 0, of the balloon ight calorimeters, allows only a partial containment of the shower. To extract optimal information we need to apply to each shower a series of cuts whose ultimate task is to separate electromagnetic and hadronic showers developing within the 4 X 0 (TS93) of the instrument. Each shower is identied by means of a few parameters based on its form and energy distribution. They can be classied as: 1. detected energy of the shower; 2. matching of the direction of the shower (given by its axis) and the extrapolation from the tracking system; 3. longitudinal prole (distribution of the energy and position of the maximum); 4. transversal prole (energy contained in a core of 3 and 1 Moliere radii); 5. starting point of the shower. To properly take into account the dependence on the energy of the electromagnetic shower the cuts imposed on these parameters have been chosen with a logarithmic dependence on rigidity. As a consequence there is a 134

137 slight dependence of the performances of the detector from the energy (TS93). In Fig the experimental distribution of a parameter related to the starting point of the shower for a) electrons and b) protons is shown. For the study of the identication procedure sample of electrons and positrons have been selected from the ight data by means of the TRD. The procedure has been weighted by means of a MonteCarlo simulation reproducing both the spectrometer and the calorimeter itself. Cut Eciency e, Contamination p Exp. data M.C. Exp. data M.C TOT Table 11.1: electron eciency and proton contamination in the TS93 calorimeter, from real data and simulation, in the rigidity range 4 to 50 GV/c. The simulator itself has been trimmed on the basis of results from tests on beams. The behaviour of electromagnetic showers of know energy was in very good agreement with the MonteCarlo results (see for instance [2]). The same code was then used to study the identication power of the TS93 Si{W calorimeter. In the Table 11.1 the electron eciency and proton contamination of each set of cuts from real data and simulation in the rigidity range 4 to 50 GV/c. Noticeable is the good agreement for electron eciency between the simulation and real data for each set of cuts and the global one while for protons there are some dierences in the eect of the various cuts still getting the same total contamination. In conclusion with the TS93 calorimeter it is possible to achieve a proton rejection factor of at of eciency CAPRICE Here we used 8 x{y Si layers and 7 X 0 of W. With an eciency of above 80% we reached, between 1 and 5 GeV, a proton contamination of to 0.01% (see Table 11.2). The analysis of the CAPRICE data is still underway and this results are preliminary. Rigidity Eciency Contamination 1{ % % 2{ % % 3{ % % Table 11.2: electron eciency and proton contamination in the CAPRICE calorimeter The Proposed PAMELA calorimeter The Pamela calorimeter takes the baseline design from the previous experiences, the characteristics needed to reach the required performances are the following: ne longitudinal segmentation: the tungsten layers have a thickness corresponding to 0.7 X 0 (0.23 cm for W); total depth: 0.9 interaction length, 16 radiation length, 23 layers; transversal segmentation type WiZard, i.e. the strip pitch is 3.6 mm; thickness of the Si detector: 380 m; the stratication is: 135

138 { Si detectors layer for the X co-ordinate, { tungsten, { Si detectors layer for the Y co-ordinate; each layer contains an array of54 Si detectors, each detector has an area of 66 cm 2 for a total sensitive area of 3024 cm 2 ; Total number of detector/plane: 542=40; number of electronic channels per x{y plane: 169=144; total number of channels: 23144=3312; total sensitive volume: cm 3 ; total volume: cm 3 ; Total mass budget: 129 Kg; Total power budget: 60 W MonteCarlo simulation of the PAMELA Calorimeter An accurate simulation of the described calorimeter has been performed by means of the GEANT code (v. 3.21). All particles are supposed to enter perpendicularly at the centre of the calorimeter (i.e. P x =0, P y =0, P z =momentum), their momentum ranging from 0.5 GeV/c to 250 GeV/c. The allowed interactions (handled by GEANT specialized data cards) are: hadronic interactions, multiple scattering, Compton scattering, pair production, bremsstrahlung production, annihilations, Landau-Vavilov energy losses (restricted), photoelectric eect, decay in ight and delta ray production. The energy thresholds for secondary productions are xed at 10 KeV. Some preliminary checks were performed in order to verify the validity of the chosen GEANT parameters. Experimental energy spectra of thin silicon detectors (300 m) associated to low noise electronics are reported in Ref [1]. The results of our simulation of a single silicon detector and the experimental spectra described in [2] are reported for comparison in Figs and Further information about the validity of both simulation and analysis procedures have been obtained by comparing our results with the experimental data gathered by the WiZard Collaboration during the test beams at CERN [2] and during the TS93 and CAPRICE 1994 balloon ights [3, 4]. Our simulated data turned out to be compatible with these experimental results. The Monte Carlo simulation has been performed for positrons, protons, electrons and antiprotons at several energies: the obtained output data have been stored in les and submitted to the analysis programs. The results included in this report are obtained with the traditional procedure that makes use of discriminating variables. We want to point out that other techniques such as cellular automata, neural networks, SPSS code have been investigated as well, giving satisfactory results in most cases. To identify the best set of discriminating variables we took advantage of our previous experience from balloon ights. Furthermore, if the eectiveness of a single variable depends also from the energy of the primary particle, good results can be obtained from a set that contains the total energy deposited in the whole calorimeter, the energy deposited in each single detector plane, the ratio between the energy deposited inside a cylinder of 4 strips of radius (4 Moliere radii) around the track and the energy deposited outside, the total number of hits (i.e. strips in which the deposited energy is higher than a xed threshold) and the shower vertex position. Samples of data at each energy are used to plot the distribution of each variable and to select the values of the cuts to be applied in order to obtain the maximum eciency with the minimum contamination (positrons over protons or antiprotons over electrons, in our case). These cuts are then applied to the whole set of data obtaining the eciency and the discriminating power of the calorimeter for each particle. The energy resolution of the calorimeter has been studied analysing a sample of positrons. The distribution of the energy deposited in the detector planes is in this case described by a gaussian. In Fig the calorimeter energy resolution as a function of the incident momentum is reported. The quoted errors are statistical. The residual contamination of protons in the positron sample is summarized in Table 11.3, together with the eciency of the calorimeter for the detection of positrons. It must be pointed out that it is possible to reach a larger proton rejection by lowering the detection eciency: this behaviour is described in Fig. 11.5, where the residual proton contamination is plotted as a function of the positron detection eciency at various energies. Finally, in Table 11.4 results obtained for the antiproton detection eciency and the residual contamination of electrons are reported. 136

139 Figure 11.2: Energy loss of a minimum ionizing particle in a silicon layer as it results from our simulation (115 GeV incident proton energy). p # # # cont. # residual contam. eciency protons positron protons positrons GeV/c % % Table 11.3: eciency of positron detection and residual contamination of protons in the positron sample. 137

140 Figure 11.3: Energy loss of a minimum ionizing particle in a silicon layer as it results from experimental spectrum (115 GeV incident proton energy) reported from [1]. Figure 11.4: Calorimeter energy resolution as a function of the incident positron momentum. 138

141 Figure 11.5: Proton contamination vs. positron detection eciency for dierent values of the incident particle momentum p # # p # cont. # residual contam. eciency electrons electrons p GeV/c % % Table 11.4: eciency of p detection and residual contamination of electrons in the p sample. 139

142 11.5 The Si detector For the proposed new stratication of the Pamela lay-out we are testing improved detectors whose characteristics are: New guard structure to minimize leakage current; The p + strip side has an n + implant with an area of 200 m 2 to allow the bias of the detector from the front; the detector has to be connected with the external world only from the side containing the strips; 7000{10000 ohm cm resistivity; higher resistivity implies lower full depletion voltage VFD, lower bias voltage; thickness 380{400 m will be established at the moment of the tender to optimize the costs; new guard structure; cut edge to allow spacers. The detectors in this design are suitable for being connected to the supporting board with a special procedure (\preform bond"). This board which contains the connecting and bias lines also supports the front end electronics and the read-out lines and is xed at his turn to the tungsten layer. Each tungsten layer is sandwiched between two detector layers, X and Y. In g we report the detailed characteristics of one PAMELA detector prototype. The current is about 1 na on all strips with a small rise for the two strips at the edge (no connection to the guard structure now oating). The preform bond is a connecting gluing procedure already widely used in the microprocessor technique and has been already used for space applications. In g 11.7 is reported a drawing of the assembly principle of the single detector in the supporting{connecting board The electronics For TS93, CAPRICE and NINA we have used a traditional surface mounted electronics. This is available but heavy, thick and large. We are proposing to make use of a VLSI chosen among the existing which are suitable for this task and have already been widely used. The chip, AMPLEX-SiCAL, is a large dynamic range front-end VLSI for silicon calorimeters designed by P. Jarron and E. Beuville, It has 16 channels, each comprising a CSA (Charge Sensitive Amplier), semi-gaussian shaper, sample-and-hold stage and output buer, in Fig 11.8 a block diagram of two channels is shown. An analogous multiplexer scans the sample-and-hold outputs. An on-chip calibration circuit, allowing electronic calibration with test pulses to 1% precision is also available as well as a fast analogue summation to compute the average input charges for trigger purposes, this would give the possibility to work in auto-trigger for calibration purposes. Other technical specications of this chip are: technology: 3 m CMOS (MIETEC); bias voltage: 5V; power consumption: 100 mw/chip. Total F.E. power consumption for the Pamela calorimeter 21 W; Peaking time: 250 ns; Cross-talk: 1.6 %; Linear dynamic range: 3.8 pc (1000 MIPs for 380 m thick detectors), ENC: 800 e, rms + 38 e, rms/pf. 18 chips are available and actually used for the construction of prototypes. This chip is operational in the luminosity monitors of the ALEPH and OPAL experiments at LEP. In order to fully optimize its performance for our application, an \ad hoc" redesigned version would be desirable. A preliminary study of the circuit parameters to be modied in the redesigned version has already been carried out. To reduce the equivalent noise charge the shaping time and the transconductance of the input transistor will be increased. This dedicated version will be produced by IMEC sharing the costs with a German group. This new version will be in the 2 m CMOS (MIETEC) technology. The nal chip will have the following improved characteristics: peaking time 1 s, noise, for 150 pf, 4000 e,. 140

143 Figure 11.6: main characteristics of one PAMELA detector prototype. Figure 11.7: drawing of the assembly principle of the single detector in the supporting{connecting board. 141

144 Figure 11.8: a block diagram of two front-end channels 142

145 11.7 The mechanics In the project of the support-container of the Pamela calorimeter we have followed design principles of : modularity; maintainability; cost containment. It should be possible to act on the detector also after the rst phase of integration. This has brought toa simple light and compact design allowing an easy interfacing with the rest of the spectrometer. The simulation and the optimization phases are completed, the response to the anticipated solicitations has been analysed with Finite Elements simulations actually. A qualication prototype is being produced. This structure, including an external support and a single detection module in scale 1:1, will allow to perform shock and vibration tests following the load and boundary conditions already used for NINA. In this way we will be able to verify the calculations which have allowed to dene all details of the mechanical design. The relevant parameters are the following: detector plane (45 silicon sensors) mm 2 ; envelope mm 3 ; total mass 130 Kg; electronics component: { components height min. 5 mm, { connectors height min. 9 mm. For the absorber we need to choose a material with highest radiation length per cm and good mechanical properties, two sintered tungsten alloys available on the market: W 97% /nickel/iron, high mechanical characteristics comparable to those of structural steel, density up to 18.5 g/cm 3 slightly magnetic (iron); W 95% /nickel/copper non magnetic more economical (easy manufacture density up to 18.1 g/cm 3. We will use 23 absorber layers (second type). The estimated absorber thickness is 2.63 mm, with 1X 0 =3.76 mm, based on the amount of W present on the alloy. The conceptual solution for the mechanical structure of the PAMELA calorimeter is the following: Figure 11.9: main structure of the calorimeter. 143

146 Figure 11.10: model used for the nite element analysis of the mechanical structure. the detection module are inserted like drawers in the main structure of the calorimeter and locked by a cover. In g this structure is presented. In g the model used for the nite element analysis is shown; the whole structure has a top ange which allows the mounting through bolts to the main plate of the experiment; each detector plane is composed of: one absorber plate and two silicon sensor planes, one for each coordinate, in the following conguration: { rst sensor plane { absorber { second sensor plane Attached to each sensor plane there is a PCB with the front end electronics; two subsequent detection planes forming a basic detector module are kept together by a frame to which they are bolted at the border of the absorber plate Fig 11.11; this frame also supports the PCBS in the part protruding outside the absorber plate; the outermost edge of this frame is glided into the supporting groves in the main structure walls; each detector module consists of two detector planes and is fully independent and extractable; in the detector module the two detector planes are bolted to the opposite sides of the support frame; the full detector has been modelled and numerically analysed by means of nite element method with operational loads as supplied by VNIIEM; each sensor plane consists of: { absorber plate 2.63 mm; { glued onto the absorber two PCB 1.6 mm; { sensors are glued on the PCB 0.38 mm; { air gap 1.0 mm; 144

147 Figure 11.11: basic module made by two detection planes corresponding to four Si layers. each PCB also supports the VLSI electronics; between PCBs of two subsequent detection planes a gap of 1.76 mm (gap between silicon 1 mm); rst and last detection planes can have dierent conguration; 23 detection planes are subdivided in 12 fully extractable and independent detection modules. The structure of the prototype to be built is made of Aluminium alloy Al 7075-T6, calculations have shown that also using this lightest material the structure supports the big mass of the absorbers and sustains the acceleration loads and has good stiness to avoid excessive displacement of planes under high loads. Figure 11.12: simulation of the stress analysis at a specic frequency. In Fig and we report a subset of the results from the simulation and optimization work. Other materials are being investigated, Titanium or steel. Titanium could represent an alternative to the expense of the mass budget and the advantage of an higher stiness, the simulation shows a sucient Margin of Safety for the Al alloy. Mechanics: mass estimate 145

148 Figure 11.13: simulated static analysis. Absorber plate 3.8 Kg 23 plates 86.5 Kg Sensor plane 0.07 Kg 46 planes 3Kg PC-Board X 0.29 Kg 23 boards 6.7 Kg PC-Board Y 0.27 Kg 23 boards 6.2 Kg Support frame 0.8 Kg 12 frames 10.2 Kg Main structure 8.2Kg 8.2 Kg Glue 0.5 Kg 0.5 Kg Electronics 8.0 Kg 8.0 Kg Total optimized Kg 146