Advanced Technology and Particle Physics

Proceedings of the 7th International Conference on

Advanced Technology & Particle Physics (IGATPP-7)

Editors M. Barone, E. Borchi, J. Huston, C. Leroy, R G. Rancoita, R Riboni & R. Ruchti

World Scientific

Proceedings of the 7th International Conference on

Advanced Technology & Particle Physics (ICATPP-7)

This page is intentionally left blank

Proceedings of the 7th International Conference on

Advanced Technology & Particle Physics (ICATPP-7) Villa Olmo, Como, Italy

15-19 October 2001

Editors

M. Barone Demokritos Laboratory, Greece

E. Borchi /A/FA/ & University of Florence, Italy

J. Huston Michigan State University, USA

C. Leroy University of Montreal, Canada

P. G. Rancoita INFN-Milan, Italy

P. Riboni ETH-Z, Switzerland

R. Ruchti University of Notre Dame, USA

l M j World Scientific « •

Sin New Jersey • London • Singapore • Hong Kong

Published by World Scientific Publishing Co. Pte. Ltd. P O Box 128, Farrer Road, Singapore 912805 USA office: Suite IB, 1060 Main Street, River Edge, NJ 07661 UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE

British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library.

ADVANCED TECHNOLOGY AND PARTICLE PHYSICS Copyright © 2002 by World Scientific Publishing Co. Pte. Ltd. All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means, electronic or mechanical, including photocopying, recording or any information storage and retrieval system now known or to be invented, without written permission from the Publisher.

For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy is not required from the publisher.

ISBN 981-238-180-5

This book is printed on acid-free paper.

Printed in Singapore by World Scientific Printers (S) Pte Ltd

Organizing Committee: M.Barone E.Borchi J.Huston C.Leroy P.G.Rancoita P.L.Riboni R.Ruchti

Democritos INFN and University of Florence Michigan State University University of Montreal INFN-Milano ETH-Zurich University of Notre Dame

International Advisory Committee: S.Baccaro B.Borgia W. B raunschweig A.Breskin A.Capone K.Freudenreich S.Giani E.Longo K.Luebelsmeyer E.Nappi A.Penzo S.Pospisil L.Price R.Reinhard L.Rossi T.J.Ruth S.Volont P.Weilhammer G.Westfall

ENEA-Rome INFN and University of Romel Aachen Weizmann Institute INFN and University of Romel ETH-Zurich CERN INFN and University of Romel Aachen INFN-Bari INFN-Trieste CTU Prague ANL ESA-Estec CERN TRIUMF ESA CERN Michigan State University

VI

Plenary Session Organizers: M.Barone (Democritos) A.Breskin (Weizmann Institute) J.Huston (Michigan) L.Price (ANL) P.L.Riboni (ETH-Zurich) T.J.Ruth (TRIUMF) S.Volont (ESA)

2nd Industry session New detectors Selected papers High Energy Physics experiments 1st Industry session Radiotherapy and Medical Imaging Space experiment

Parallel Session Organizers: S.Baccaro (ENEA, Rome) W.Braunschweig (Aachen) A.Breskin (Weizmann Institute) A.Capone (Romel) K.Freudenreich (ETHZ) S.Giani (CERN) C.Leroy (University of Montreal) E.Longo (Romel) E.Nappi (INFN-Bari) S.Pospisil (CTU, Prague) L.Price (ANL) R.Reinhard (ESA-Estec) R.Rutchi (Univiversity of Notre Dame) T.J.Ruth (TRIUMF) P.Weilhammer (CERN) G.Westfall (Michigan State University)

Secretariat: C. Dolfi N.Tansini

Florence Villa Olmo

Radiation Damage (both sessions) Tracker with non Si Substrate New detectors Passive Particle Physics Silicon Tracker II and III Software applications Calorimetry (both sessions) Crystal Detectors Particle Identification Medical Application Instrumentation High Energy Physics Experiments Space Experiments Poster session Radiotherapy and Medical Imaging (both sessions) Silicon Tracker I Medium Energy Physics

VII

PREFACE The exploration of the subnuclear world is done today through increasingly complex experiments covering a wide range of energy and performed in a large variety of environments going from particle accelerators, underground detectors up to satellites and space laboratory. The achievement of these research programs calls for novel techniques, new materials and new instrumentation to be used in detectors, often of large scale. Therefore, particle physics is at the forefront of technological advance and also leads to many applications. Among these, medical applications have a particular importance due to health and social benefits they bring to the public. The International Conference on Advanced Technology and Particle Physics was aimed at reviewing the advances made in all technological aspects of the experiments at various stages, running, upgrade or in preparation. The open and flexible format of the Conference was conducive to fruitful exchanges of points of view among participants. The medical sessions gave an interesting example of merging advanced technology, particle physics and numerical techniques. Industries specialized in advanced technologies were present at the Conference through two dedicated plenary sessions and a show of products of industry. Plenary and parallel sessions covered Space and Astroparticle Physics experiments (organized by R. Reinhard, S. Volonte), Silicon Tracker (K. Freudenreich, P. Weilhammer, W. Braunschweig), Medium Energy Experiments (G. Westfall), Calorimetry (C. Leroy), Radiation Damage (S. Baccaro), Passive Physics Experiments (A. Capone), Radiotherapy and Medical Imaging (T. Ruth) and Medical Application Instrumentation (S. Pospisil), Software Applications (S. Giani), Particle Identification (E. Nappi), High Energy Physics Experiments (L. Price), Industry (M. Barone, P.L. Riboni), New Detectors (A. Breskin), Crystal Detectors (E. Longo), Poster Session (R. Ruchti), Selected Papers (J. Huston). Several sessions have been merged for the proceedings edition. The Conference welcomed about 250 participants in the very pleasant "Centro di Cultura Scientifica A. Volta". These participants were representing 80 institutions from 20 countries. We would like to thank the staff of Centro A. Volta for the excellent support provided to the Conference organization. In particular, we would like to extend our appreciation and thanks to the Secretariat of Centro di Cultura Scientifica A. Volta for their help and efficiency with the organization of the Conference and its running. The help of Celine Lebel, from Montreal University, in the preparation of the Conference proceedings is gratefully acknowledged.

viii The organizers would like to thank deeply the strong support of INFN and ESA which made the conference possible. Finally, we would like to thank the speakers for the high quality of their contributions and the participants for their enthusiasm in attending the Conference and contributing to the discussions.

Michele Barone Emilio Borchi Joey Huston Claude Leroy Pier-Giorgio Rancoita Pier-luigi Riboni Randal Ruchti March 2001

ix CONTENTS

Preface

Space and Astroparticle Physics Experiments The Alpha Magnetic Spectrometer, a Particle Physics Experiment in Space HYPER: Atom Interferometry in Space The GLAST Gamma ray Large Area Telescope The Anticoincidence Shield of the PAMELA Satellite Experiment . . EUSO: Extreme Universe Space Observatory A Complete Simulation of Cosmic Rays Access to a Space Station . The Time of Flight System of the AMS-02 Space Experiment . . . . The Transition Radiation Detector of the AMS-2 Experiment . . . . Test of a Novel Detector Technique: the ICARUS T600 Module . . . MONOLITH: a Massive Magnetized Detector for Atmospheric Neutrinos Development and Performances of the MAGIC Telescope New Developments of Photodetectors for the Lake Baikal Neutrino Experiment Operation and Calibration of Large-mass Droplet Detectors for PICASSO GLAST Tracker Tray: Thermal and Dynamic Performance

vii

1 3 16 23 31 36 45 52 57 62 68 74 79 85 95

Silicon Tracker

101

Convener's Report K. Freudenreich Design of an Upgraded D 0 Silicon Microstrip Tracker for Fermilab Run2B DO Silicon Microstrip Tracker for Run IIA The CDFII Silicon Tracking System The CDF Online Silicon Vertex Tracker Commissioning and Operation of the CDF Silicon Detector The Assembly of the AMS Silicon Tracker, Version 1 and 2 The AMS Infrared Tracker Alignment System - from STS91 to ISS Performance of the BaBar Silicon Vertex Tracker Charged Particle Tracking with the HERA-B Detector The ZEUS Micro Vertex Detector The Run IIB Upgrade of the CDF Silicon Detectors

103

110 115 120 127 134 143 149 154 159 166 171

X

The BTeV Pixel Detector System Monolithic CMOS Pixels for Charged Particle Tracking Status and New Layout of the ATLAS Pixel Detector The ATLAS Silicon Microstrip Tracker Construction Status The Silicon Strip Tracker of the CMS Experiment The CMS Silicon Tracker Automated Module Assembly CMS Silicon Tracker - Milestone 200 Test of the CMS Silicon Strip Detectors in the Hadron Beam . . . . Status of the CMS Pixel Detector Fabrication of Microstrip Detectors and Integrated Electronics on High Resistivity Silicon The Diamond Project at GSI - Perspectives Radical Beam Gettering Epitaxy of ZnO and GaN GEM Detectors for COMPASS Architecture of the Common Gem and Silicon Readout for the COMPASS Experiment

178 183 189 196 203 209 219 224 231

Medium and High Energy Physics Experiments

269

Performance of the Pre-shower System in the HADES Spectrometer The Time Projection Chamber for the CERN-LHC Heavy-ion Experiment ALICE Cathode Strip Chambers Data Analysis A Gas System for a Large Multi-cells Detector Run II Upgrades and Physics Prospects Detectors for a Linear Collider The ATLAS Muon Spectrometer US ATLAS Muon End Cap System Performance of the MACRO Limited Streamer Tubes for Estimates of Muon Energy Exploitation of ATLAS DAQ Prototypes for Test Beam and Lab Activities Cathode Strip Chamber Performance of the CMS ME1/1 Muon Station The Run2 D0Muon System at the Fermilab Tevatron The D 0 Central Tracker Trigger A Proposal for the Alignment of the LHCB RICH Detector Monitored Drift Tube Chamber Production at Laboratori Nazionali di Frascati A Database for Detector Conditions Data of Current and Future HEP Experiments

271

241 246 252 259 264

276 282 289 300 309 320 327 332 339 347 352 357 363 368 373

xi

Calorimetry

379

Convener's Report C. Leroy Overview of the CMS Electromagnetic Calorimeter The Readout of the ATLAS Liquid Argon Calorimeter A New W/Scintillator Electromagnetic Calorimeter for ZEUS . . . Performance of the ATLAS Liquid Argon Electromagnetic Calorimeter Modules under Test Beam Status of ATLAS Tile Calorimeter and Study of Muon Interactions . Construction of the First CMS-EC AL Fully Operational Module (400 Lead Tungstate Crystals) A New Concept for an Active Element for the Large Cosmic Ray Calorimeter ANI What's New with the CMS Hadron Calorimeter Overview of the ATLAS Liquid Argon Calorimeter System The ATLAS Hadronic Endcap Calorimeter

381

Radiotherapy and Medical Imaging

465

Progress of Heavy Ion Therapy Inorganic Scintillators for Medical Diagnostics A Solution for Dosimetry and Quality Assurance in IMRT and Hadrontherapy: the Pixel Ionization Chamber Biological Interpretation of Quantitative P E T Brain Data Characterization of the BNCT Epithermal Column of the Fast Reactor Tapiro (ENEA) and Dose Measurements in Phantom Utilising Not-conventional Detection Radioactive Ion Beams for Bio-medical Research and Nuclear Medical Application Production of Radioisotopes for Imaging and Therapy at Low Energy

467 474

Technology Transfer and Education

519

Industrial Section Convenor's Report M. Barone Electronic Publishing at the End of 2001 DAQ Cards for the Compact Muon Solenoid: a Successful Technology Transfer Case The Data Acquisition System for the CMS Experiment at the LHC . Quarknet: a Particle Physics Program of Education and Outreach in the U.S.A

521

384 389 396 401 409 417 424 429 435 449

487 492

499 504 512

525 534 540 550

xii

Particle Identification Development of High Time Resolution Multigap RPCs for the TOF Detector of Alice The LHCb Ring Imaging Cherenkov Detectors Particle Identification with the HERA-B RICH A RICH Detector for Hadron Identification at Jefferson Lab, Hall A The Silicon Transition Radiation Detector: a Test with a Beam of Particles The ALICE Transitic Radiation Detector: Results from Prototype Tests The Silicon Transition Radiation Detector: a Full Monte Carlo Simulation

561 563 569 576 581 587 592 597

N e w Detectors

603

Gravitational Waves Interferometers and the VIRGO Project . . . . New Developments in the Position Sensitive Detectors based on MicroChannel Plates Further Studies of the Sand-Glass Gas (SGG) Detector Registration of Charged Particles by Scintillating Fibers coupled with //-cell SI APD G The Micromegas Neutron Detector for CERN N_TOF Scintillator-fiber-based Inner Tracking Detectors for the D 0 Experiment at Fermilab Tests of RPCs for the ARGO Experiment at YB J Large Liquid Scintillator Tracker for Neutrino Experiments Low-temperature Thermal Characterization of Support Material for Massive Cryogenic Detectors Development of Ti based Transition Edge Sensors for Cryogenic Detectors Measurement of Electron-phonon Decoupling in NTD31 Germanium A Study of Micromegas with Preamplification with a single GEM . . The Antiseismic Suspension for the VIRGO Project CMOS Circuits to drive QW Modulators The Advanced Study of Silicon Photomultiplier

605

Crystal Detectors

729

Convener's Report E. Longo Lead Tungstate Crystals for the CMS Electromagnetic Calorimeter . Increase in Photon Collection from a YAP:CE Matrix coupled to Wavelength Shifting Fibres Development of 300 g Scintillating Calorimeters for WIMP Searches

731

615 620 627 633 639 644 661 668 677 684 694 704 712 717

735 740 746

xiii

Scintillators for Photon Detection at Medium Energies First Experiences with the Mainz Lead Fluoride Calorimeter

....

751 758

Radiation Damage

765

Convener's Report S. Baccaro Radiation Effects in Silicon Detectors: a Short Overview Study of Radiation (Neutron, 7-ray, and Carbon-iron) Effects on NPN Bipolar Transistors Radiation Tolerance of a 0.18 /im CMOS Process Total Dose Test for Commercial Off-the-shelf Components to be used in a Space Experiment: a Survey on Current Technologies Liquid Argon Pollution Tests of ATLAS Detector Materials at the IBR-2 Reactor in Dubna Irradiation Test of the ZEUS Vertex Detector Frontend Chips, the Helixl28-3.0 Radiation induced Color Centers in Tb 3 + -doped Phosphate Scintillation Glasses Point Defects in Lithium Fluoride Films Induced by Gamma Irradiation In Situ Measurement of Radiation Damage in Scintillating Fibers . . Irradiation Effects on Poly (Vinyl Chloride) Influence of the Neutron-caused Defects on the Parameters of Magnetic Microsensors and Methods for Improvement of their Radiation Hardness

767

List of participants

770 780 787 792 800 806 811 819 826 831

836

843

Space and Astroparticle Physics Experiments Organizers: A. Capone (Passive Particle Physics) R. Reinhard (Space Experiments-Parallel) S. Volonte (Space Experiments-Plenary) R. Battiston A. Landragin S. Brez J. Lund L. Scarsi D. Grandi C. Sbarra G. Schwering C. Vignoli G. C. Trinchero D. Bastieri B. K. Lubsandorzhiev C. Leroy M. N. Mazziotta

The Alpha Magnetic Spectrometer, a Particle Physics Experiment in Space HYPER: Atom Interferometry in Space The GLAST Gamma ray Large Area Telescope The Anticoincidence Shield of the PAMELA Satellite Experiment EUSO: Extreme Universe Space Observatory A Complete Simulation of Cosmic Rays Access to a Space Station The Time of Flight System of the AMS-02 Space Experiment The Transition Radiation Detector of the AMS-2 Experiment Test of a Novel Detector Technique: the ICARUS T600 Module MONOLITH: a Massive Magnetized Detector for Atmospheric Neutrinos Development and Performances of the MAGIC Telescope New Developments of Photodetectors for the Lake Baikal Neutrino Experiment Operation and Calibration of Large-mass Droplet Detectors for PICASSO GLAST Tracker Tray: Thermal and Dynamic Performance

This page is intentionally left blank

T H E A L P H A M A G N E T I C S P E C T R O M E T E R , A PARTICLE PHYSICS E X P E R I M E N T I N SPACE R. B A T T I S T O N Dipartimento

di Fisica dell' Universitd and Sezione INFN, Perugia, Italy E-mail: [email protected]

Via Pascoli,

06100

The Alpha Magnetic Spectrometer (AMS) is a state of the art detector for the extraterrestrial study of matter, antimatter and missing matter. During the STS91 precursor flight in may 1998 AMS collected nearly 100 millions of Cosmic Rays on Low Earth Orbit, measuring with high accuracy their composition. We present results on the flux of proton, electron, positron and helium. Analyisis of the under cutoff spectra indicates the existence of a new type of belts of energetic trapped particles characterized by a dominance of positrons versus electrons.

1

Introduction

The disappearence of the antimatter 1 ' 2 ' 3 and the presence at all scales in our universe of a non luminous components of matter (dark matter) 4 ' 5 are two of the most intriguing misteries in our current understanding of the structure of the Universe. To study these problems, a high energy physics experiment, the Alpha Magnetic Spectrometer (AMS) 6 , is scheduled for installation on the International Space Station in 2004. Goal of AMS is to perform a three year long measurement, with the highest accuracy, of the composition of Cosmic Rays in the rigidity range 0,1 GV to several TV. In preparation for this long duration mission AMS flew a ten days precursor mission on board of the space shuttle Discovery mission STS-91 in June 1998. This high statistics measurement of CR in space, enabled, for the first time, the systematic study the behaviour of primary CR near Earth in the rigidity interval from 0,1 GV to 200 GV, at all longitudes and latitudes up to ±51.7°. In this paper we present some relevant results obtained by AMS during the precursor mission. We also report the observation of high energy radiation belts in the near Earth region and on their composition, which shows remarkable differences with previously observed belts of trapped particles around our planet.

3

4

Figure 1. A MS on the Discovery STS 91 precursor flight, June 1998.

2

The A M S experiment on the STS-91 mission

Search of antimatter requires the capability to identify with the highest degree of confidence, the mass of particle traversing the experiment together with the absolute value and the sign of its electric charge. The AMS configuration flown in 1998 on the Shuttle Discovery (Fig.l) includes a permanent Magnet, Anticounter (ACC) and Time of Flight (ToF) scintillator systems, a large area, high accuracy Silicon Tracker and an Areogel Threshold Cherenkov counter. The magnet is based on recent advancements in permanent magnetic material and technology which make it possible to use very high grade Nd-Fe-B to build a permanent magnet with BL2 = 0.15 Tm2 weighting < 2 tons. A charged particle traversing the spectrometer triggers the experiment through the ToF system, which measures the particle velocity with a resolution of ~ 120 ps over a distance of ~ 1.4 m 1 1 . The pattern recognition and tracking is performed using the large area

5 1

Iff

D 0.2 < 6 M

< 10"8 AMS on ISS (2003-6)

10 s

10"'

10'1

-I

• ' '""'

I

1

• '

I 10

I . 10s

'

• • • •••"

10'

Rigidity (GV)

Figure 4. Antimatter limits.

large He sample collected by AMS a search for anti-He candidates has also been performed. Within 2.3 Millions He events no anti-He candidates have been found, up to a rigidity of 140 GV. Assuming identical He and anti-He spectra we obtain a model independent upper limit of 1.110 -6 over the rigidity interval 1 to 140 GV, which can be compared to previous results (Fig.4). 3

Observation of high energy particle belts

The trapping of charged particles in the quasi dipolar earth magnetic field is a classical problem, which has been studied in great detail 27 following Van Allen observations in 1958 29 . The basic physical mechanism is well understood. For sufficiently low rigidities, the trapped particles spiralize along orbits defining shells surrounding our planet. These shells are shaped along the magnetic field lines and are roughly

9 7

L _ -

b

5 r

-

AM8 BESS98 CAPR1CE94 IMAX92 MASS91 LEAP87

iiittt-%5

Safe

J

4 3

7

2

-

-1 0 -1

TUL A

(a)

A°

* *

9*

~m*St r.

. q

1«2

Figure 5. (a)Primary proton flux measured by AMS and compared with existing balloons measurements. The lines are parametrizations of the primary cosmic rays used in atmospheric u flux calculation: dashed line HPPK 2 3 , dot-dashed line Bartol group 2 4 ; (b) primary He flux measured by AMS and compared with existing balloons measurements.

symmetric in latitude with respect to the geomagnetic equator. The motion of a trapped particle can be separated in three components, the revolution around the guiding center or gyration, the bouncing between mirror points located « symmetrically with respect to the geomagnetic equator (magnetic bottle), and a longitudinal drift around the earth. The geometrical locations defined by the orbits of trapped particles are called shells.A shell can be univocally determined by two parameters. For example a pair of variables are L, the distance of the shell at the equator measured in unit of the Earth radius (i?©), and -B m , r , the value of the magnetic field at the point where the particles reverse their motion (mirror point) 30 . Depending on the shell, Bmir can be locally very deep the atmosphere (it can be below the earth

10

crust). Shells which are characterized by these value of BmiT cannot trap the particles, since they are lost within one or few bounces across the magnetic equator. A particle belonging to a shell will remain on the same shell until it is disturbed by (a) interaction with the top layers of the atmosphere or other particles or (b) interaction with electrical or magnetic variable field. Conversely, primary cosmic rays coming from deep space cannot enter a shell unless their trajectories are disturbed by some interaction with matter or fields. The existence of the shells is the result of the equilibrium between two mechanisms: some contributing to fill the shells with new particles and others removing some of the trapped particles. If the dynamics of the particles trapped is well understood, the mechanisms contributing to shell stability are much less understood. They involve: interaction of high energy CR with the atmosphere creating neutrons which decays in flight, n —>• p + e~ + Ve + 782KeV, filling the belts (CRAND mechanism 28 ), instabilites due to solar storms, as well as other types of magnetic and electric instabilities. It should be pointed out, however, that the mechanisms proposed are compatible with the observed dominance of protons and electrons in the Van Allen belts. The shell can be classified by their composition and location. The original Van Allen belts contain only proton and electrons and extend to very large distance from the earth, up to L « 6. Van Allen belts are divided into inner and outer belts, since there is a dip in the particle flux intensitiy at about 2.5 L. During the last 20 years, there have been reports of the observation of a low flux of trapped ions, mainly He and O, with traces of C e N, and having energies of a few MeV/n and L — 3 — 4. These particles are extracted from the upper layers of the atmosphere during solar storms. More recently, nearly 40 years after Van Allen discovery, the analysis of SAMPEX data 32 has shown the existence of belts included in the inner Van Allen belts, containing heavier nuclei like N, O, Ne with rigidities of the order of 10/MeV. The SAMPEX belts are different from the Van Allen belts mainly because of their composition due to a different the filling mechanism, which is likely due to the interaction of the so called Anomalous Cosmic Rays with the Earth atmosphere 33,31 .

11 Table 2. Different types of particle belts around the Earth.

Belt type Van Allen (inner) Van Allen (outer) SAMPEX AMS

Particle type P e~ e~ P N+x,0+x, Ne+X P e~ e+ 3 He

Rigidity

[MeV/n] 0.1 - 100 0.01 - 1 1-10 0.1-1 10 10 - 100 100 - 1000 100 - 1000 100 - 1000 100 - 1000

Filling mechanisms n -*• pe~Ve, external belts solar wind Anomalous CR primary CR interacting with the atmosphere

L 2.5

1-10

2

10 - 100

< 1.15

io- B - io~ 4

The belts observed by AMS are different in composition since they also contain a large fraction of positrons, but also deuterium and 3He. These particles have not been observed in the Van Allen or SAMPEX belts. Particularly striking is the abundance of positrons versus electrons, with a ratio exceeding a factor of four in the equatorial region. AMS observed shells with L < 1.15, well below the inner Van Allen belts. In the belts studied by AMS the observed proton spectrum is harder than in the case of Van Allen belts. This can be understood since their location is closer to the earth and the particles do experience a stronger trapping field. Another difference with the Van Allen belts is the residence time of the trapped particles, computed using computer based tracing techniques, which is in the region of seconds and not days or weeks. These shells cannot be observed by stratospheric balloons, since their mirror fields are above the atmosphere except in correspondence of the South Atlantic Anomaly. It follows that the observed particles do not belong to the various types of albedo particles reported in the past by experiments on balloons. In Table 2 we summarize the main features of the different type of belts identified during the last 40 years. As we can see the situation is very varied, corresponding to different filling mechanisms. Since we are dealing with continuous distributions, the reported intervals (rigidity, L, residence time)

12 Table 3. Physics capabilities of AMS after three years on the ISS

Elements e+ P 7 He/He C/C D,H2 3 He/ 4He 10 Be/ 9Be

Sensitivity 108 500000 l

W

9 10 To* 10 9 2%

(Now)

(~ 10a) (-30)

(if)

(A)

Energy Range(GV) 0.1 - 100 0.5 - 100 0.1 - 300 0.5 - 20 0.5 - 20 1.0-3.0 1.0-3.0 1.0-3.0

Physics

t Dark Matter

I Antimatter CP, GUT, EW

T Astrophysics

4

separatee1 from p,p upto300GeV detect v->e + e~ pairs up to 100 GeV

LEPS

He3,He4,B,C,... e'.ytolOOOGeV

Figure 6. Configuration in S004.

of AMS on the ISS for the three years mission scheduled on UF4

should be taken as typical order of magnitudes.

13

4

Conclusions

The first mission of the Alpha Magnetic Spectrometer, although lasting only ten days, has been scientifically very rewarding, allowing for the first time a very detailed measurement of high energy cosmic rays outside the atmosphere. In addition to the most accurate measurements obtained so far for the primary flux of p, e + , e~, D,3 He and 4He spectra over most of the earth surface, these results have shown the existence of a substantial second spectrum of high energy particles trapped within low altitude belts. These new belts have a very characteristic composition, dominated by positively charged particles, mainly p, e+ and D. Their existence should be taken into account when calculating radiation doses for astronauts on the ISS or background rates for low orbit satellites. AMS is currently being refurbished to be ready for a three years mission with UF4 in 2004.A stronger magnetic field from a superconducting magnet, B — 0,7 T, a fully equipped Silicon Tracker, together with three powerful particle identification detectors, a Transition Radiation Detector, a Ring Imaging Cherenkov (RICH) detector and an Electromagnetic Calorimeter, will allow precise particle identification up to 0(TeV) of energy (Fig.6). The physics capabilities of AMS after three years of exposure on the ISS are summarized in Table 3. AMS will be the only large acceptance magnetic facility which will be exposed for long time in space. It will allow a measurements of the flux and composition of Cosmic Rays with an accuracy orders of magnitude better than before. The large improvement in sensitivity given by this new instrument, will allow us to enter into a totally new domain to explore the unknown. 5

Acknowledgment

This work has been partially supported by the Italian Space Agency (ASI) under contract ARS-98/47. References 1. Steigmann, G., Ann. Rev. Astron. Astroph., 14 p. 339, 1976.

14

2. Kolb, E.W., Turner, M.S., Ann. Rev. Nucl. Part. Sci. 33 p. 645, 1983. 3. Peebles, P.J.E., Principles of Physical Cosmology, Princeton University Press, Princeton N.J., 1993. 4. Ellis, J. et al., Phys. Lett. B214, p. 403, 1988. 5. Turner, M.S., Wilzek, F., Phys. Rev. D42, p. 1001, 1990. 6. Ahlen, S. et al., Nucl. Inst. Meth. A350, p. 351, 1994. 7. Battiston, R., Nucl. Instr. Meth. (Proc. Suppl.) B44, p. 274, 1995. 8. Acciarri, M. et al., Nucl. Inst. Meth. A289 p. 351-355, 1990. 9. Alcaraz, J. et al., II Nuovo Cimento 112A, p. 1325-1344, 1999. 10. Batignani, G. et al., Nucl. Inst. Meth. A277 p. 147, 1989. 11. Alvisi, D. et al., Nucl. Inst. Meth. A437 p. 212, 1999. 12. Produced at CSEM, SA Rue J. Duot 1, P.O. Box, CH-2007 Neuchatel, http://www.csem.ch. 13. AMS Collaboration, Alcaraz, J. et. al., Search for Antihelium in Cosmic Rays, Phys. Lett. B461, p. 287, 2000. 14. AMS Collaboration, Alcaraz, J. et. al., Protons in Near Earth Orbit, Phys. Lett. B472, p. 215, 2000. 15. AMS Collaboration, Alcaraz, J. et. al., Leptons in Near Earth Orbit, Phys. Lett. B484, p. 10, 2000. 16. AMS Collaboration, Alcaraz, J. et. al., Cosmic Protons, Phys. Lett. B490, p. 27, 2000. 17. Lamanna, G. PhD Thesis, University of Perugia, October 2000, unpublished. 18. BESS98, Sanuki, T. et al., astro-ph/0002481, 2000. 19. CAPRICE94, Boezio, M. et al., ApJ 518, p. 457, 1999. 20. IMAX92, Menn, W. et al., The Astroph. J. 533, p. 281, 2000. 21. MASS91, Bellotti, R. et al., Phys. Rev. D60, p. 052002, 1999. 22. LEAP87, Seo, E.S. et al., ApJ 378, p. 763, 1991. 23. HPPK, Honda, M. et al., Phys. Rev. 52, p. 4985, 1995. 24. BARTOL, Lipari, P. and Stanev, T., Talk given at Now 2000 Conference, 2000. 25. Smoot, G.F. et al., Phys. Rev. Lett. 35 p. 258-261, 1975; Steigman, G. et al., Ann. Rev. Astr. Ap. 14 p.399, 1976; Badhwar, G. et al., Nature 274 p. 137, 1978; Bufnngton, A. et al., ApJ 248 p. 1179-1193,

15

1981; Golden, R.L. et al., ApJ 479 p. 992,1997; Ormes, J.F. et al., ApJL 482 p. 187, 1997; Saeki,T. et al., Phys. Lett. B422 p. 319, 1998. 26. Nozaki, M., OG.1.1.23, 26th ICRC, Salt Lake City, Utah, 1999. 27. For a recent review see Walt,M., Radiation Belts Models and Standards. AGU Geophysical Monograph 97, p.l, 1997. 28. Singer,S.F., Phys. Rev. Lett. 1 p. 181, 1958; Hess, W.N., Phys. Rev. Lett.3 p . l l , 1959; Kellogg, P., J.Geophys. Res. 65 p. 2,705, 1960; Vernov, S.N. et al., Soviet Physics, Doklady 4 p.154, 1959. 29. Van Allen, Ludwig, Ray, and Mcllwain, IGY Satellite Series Number 3, 73 Natl. Acad. Sci., Washington D.C., 1958; Van Allen, Mcllwain, and Ludwig, J. Geophys. Research 64, p. 271, 1959; Van Allen, J.A. and Frank. L.A., Nature 183 p. 430, 1959. 30. Mcllwain, C.E., J.Geophys. Res. 66p. 3681-3691, 1961. 31. Mewaldt, R.A., Radiation Belts Models and Standards. AGU Geophysical Monograph 97, p.35, 1997. 32. Cook, W.R., IEEE Trans. Geosci. Remote Sensing 31 p. 557-564, 1993. 33. Cummings, J.R. et al., Geophys. Res. Lett. 20 p. 2003-2006, 1993; Cummings, J.R. et al., IEEE Trans. Nucl. Sci. 40 p. 1459-1462, 1993.

H Y P E R : ATOM I N T E R F E R O M E T R Y I N SPACE

A. LANDRAGIN, A. CLAIRON, N. DIMARCQ, P. TEYSSANDIER, C. SALOMON SYstemes de Reference Temps-Espace, Observatoire de PARIS, France E.M. RASEL, W. ERTMER Institut fur Quantenoptik, Universitat Hannover,

Germany

CH.J. BORDE, P. TOURRENC Universite Pierre et Marie Curie, Equipe de Relativite, Gravitation et Astrophysique, France P. BOUYER Laboratoire Charles Fabry de I'Institut d'Optique, France M. CALDWELL, R. BINGHAM, B. KENT, M. SANDFORD Rutherford Appleton Laboratory, UK England

Bureau International

P. WOLF des Poids et Mesures, Sevres, France

S. AIREY, G. BAGNASCO ESTEC, ESA, The Netherlands The objective of the HYPER project is to use the very high sensitivity of the atomic interferometry in space for research in fundamental physics This project is sustained by many scientists in the atomic physics community x and by ESA. After a short introduction, the second part will describe the recent development in atom interferometry. The third part is dedicated to the description of the mapping of the Lense-Thirring effect with an atomic gyroscope.

1

Introduction

Inertial Sensors are useful device in both science and industry. Higher precision sensors could find practical scientific applications in the areas of general relativity 2 , geodesy and geology. Important applications of such devices occur also in the field of navigation, surveying and analysis of structures. Matterwave interferometry has recently shown its potential to be an extremely sensitive probe for inertial forces 3 . First, neutron interferometers have been used to measure the rotation of the earth 4 and the acceleration due to gravity 5

16

17

in the end of the seventies. In 1991, atom interference techniques have been used in proof-of-principle work to measure rotations 6 and accelerations 7 . In the following years, many theoretical and experimental works have been performed to investigate this new kind of inertial sensors 8 . Some of the recent works have shown very promising results leading to sensitivity comparable to other kind of sensors, as well for rotation 9 as for acceleration 10 . 2

Inertial sensors based on atom interferometer: basic principal

We present here a summary of recent work with light-pulse interferometerbased inertial sensors. We first outline the general principles of operation of light-pulse interferometers. This atomic state interferometer n ' 1 2 uses twophoton velocity selective Raman transitions 13 , to manipulate atoms while keeping them in long-lived ground states. 2.1

Principle of a light pulse matter-wave

interferometer

Light-pulse interferometers work on the principle that when an atom absorbs or emits a photon momentum must be conserved between the atom and the light field. Consequently, an atom which emits (absorbs) a photon of momentum hkeff will receive a momentum impulse of dp = —hkeff(+hkeff) . When a resonant traveling wave is used to excite the atom, the internal state of the atom becomes correlated with its momentum: an atom is in its ground state |1) with momentum p (labeled |l,p) ) is coupled to an excited state |2) of momentum (\2,p + hkeff)) n . A precise control of the light-pulse duration allows a complete transfer from one state (for example |l,p)) to the other (\2,p + hkeff)) in the case of a n pulse and a 50/50 splitting between the 2 states in the case of a n/2 pulse (half the duration of a 7r pulse). We use a n/2 — n — TT/2 pulse sequence to coherently divide, deflect and finally recombine an atomic wavepacket. The first n/2 pulse excites an atom initially in the \l,p) state into a coherent superposition of states |l,p) and |2,p + hkeff). If state |2) is stable against spontaneous decay, the two wavepackets will drift apart by a distance hkT/m in time T. Each wavepacket is redirected by a n pulse which induces the transitions |l,p) —• \2,p + hkeff) and \2,p+hkeff) -> |l,p). After another interval T the wavepacket once again overlap. A final pulse causes the two wavepackets to interfere. The interference is detected, for example, by measuring the number of atoms in the |2) state. We obtain large wavepacket separation by using laser cooled atoms and velocity sensitive stimulated Raman transitions 13 to drive the transitions.

18

2.2 Application to Earth-based inertial sensors Inertia! forces manifest themselves by changing the relative phase of the de Broglie matter waves with respect to the phase of the driving light field, which is anchored to the local reference frame. The physical manifestation of the phase shift is a change in the number of atoms in, for example, the state |2), after the interferometer pulse sequence as described above. If the 3 light pulses of the pulse sequence are only separated in time, and not separated in space {i.e. if the velocity of the atoms is parallel to the laser beams), the interferometer is in a gravimeter or accelerometer configuration . In a uniformly accelerating frame with the atoms, the frequency of the driving laser changes linearly with time at the rate of —keff.a. The phase shift arises from the interaction between the light and the atoms 8 and can be written: A^ =

fc(ti)-2fc(ta)

+ &(t 8 )

(1)

where fcfa) is the phase of light pulse t at time U relative to the atoms. If the laser beams are vertical, the gravitationally induced chirp can be written: A4> = ~keff.gT2

(2)

Figure 1. Principle of the atom-fountain-based atom gravimeter achieved in S. Chu (Nobel 1998) group at Stanford. Left shows a two days recording showing the variation of gravity (top curve). The accuracy enable to resolve ocean loading effects (curve i and ii correspond to residual compare to models with or without ocean loading taken into account).

19

Rotation rate (xlO° rad.s"1)

Figure 2. Schematic of the atomic Sagnac interferometer at Yale 9 on left. Individual signals from the outputs of the two interferometer(gray lines), and difference of the two signals corresponding to a pure rotation signal (black line) versus rotation rate.

It should be noted that this phase shift does not depend on the atomic initial velocity or on the mass of the particle. Recently, atomic gravimeter with comparable accuracy than best comer cube has been achieved 10 . The main limitation of this kind of gravimeter on earth is is due to spurious acceleration from the reference platform. Measuring gravity gradient may allow to overcome this problem, indeed, by using the same reference platform for two independent gavimeters enable to extract gravity Iuctuations. Such apparatus 14, using two gravimeters as described above but shearing the same light pulses, has shown a sensitivity of 3.10 _8 s -2 .i?2; -1 / 2 and as a potentiel on earth up to 10~9s~2.Hz~1^2. If the laser beams are separated in space (i.e. if the atomic velocity is perpendicular to the direction of the laser beams), the interferometer which is formed is in a Mach-Zenhder configuration. In this case, the interferometer is also sensitive to rotations, as in the Sagnac geometry 15 for light interferometers . For a Sagnac loop enclosing area A, a rotation Q produces a phase shift: 4w ^V Sagnac

=

T

".A

(3)

where A is the particle wavelength and «£ its longitudinal velocity. The area A of the interferometer depends on the distance between two pulses L and recoil velocity Vr = %kjm:

A-L2^ A

L

~~ vL

(4)

20

Thanks to the use of massive particle, atomic interferometer can achieve very high sensitivity. An atomic gyroscope 9 using thermal caesium atomic beams {VL ~ 300ra.s _1 ) and with a overall interferometer length of 2m has demonstrated a sensitivity of 6.10~10rad.s~1.Hz~1/2. The apparatus consists on a double interferometer using two counter-propagating sources of atoms and shearing the same lasers which enables to discriminate between rotation and acceleration. Improvements of an atomic Sagnac interferometer relies on the increase of the enclosed surface which is determined by the ratio of the atomic beam velocity vi to the velocity VT of both atomic waves relative to each other due to the beam splitting process. Therefore using cold atomic source (with velocity dispersion close to the recoil velocity) will enable to achieve a ratio of VT/VL close to unity. The improvements with HYPER will follow precisely this philosophy and will benefit from the space environment, which enable very long interaction time (few seconds) and low spurious vibrational level. Presently first prototypes based on atomic fountains of laser cooled atoms are under construction in a joint project of LPTF, IOTA and LHA in Paris as well as at the IQO in Hannover. 3

Latitudinal mapping of the gravitomagnetic effect with HYPER

The Lense-Thirring effect consists on a precession of a local referential frame (realized by inertial gyroscopes) and non-local referential realized by direction of fixed stars. This Lenth-Thirring precession is given by: nLT=

~j

GIZ{Lj.r)r-wr2 ~5

W

The high sensitivity of atomic Sagnac interferometer for rotation rates will enable HYPER to measure the latitudinal structure of the gravitomagnetic or Lense-Thirring effect while the satellite orbits around the Earth. In a Sunsynchronous, circular orbit at 700 km altitude, HYPER will detect how the direction of the Earth's drag varies over the course of the near-polar orbit as a function of the latitudinal position B: OA 3 / sin(20) ny )(X2\cos{2B)-\ with 7 | | e ^ , 9 = arcos(r.ex) defines the orbital plane.

\ )

(6)

the coordinate system, spanned by ex and ey,

21

Figure 3. Schematic of the measurement of the Lense-Thirring effect. The black lines visualise the vector field of the Earth's drag SILT- The sensitive axes of the two ASUs are perpendicular to the pointing of the telescope. The direction of the Earth's drag varies over the course of the orbit showing the same structure as the field of a magnetic dipole. Due to this formal similarity the Lense-Thirring effect is also called gravitomagnetic effect. The modulation of the rotation rate QLT due to Earth's gravitomagnetism as sensed by the two orthogonal ASUs in the orbit around the Earth appears at twice the orbit frequency.

HYPER carries two atomic Sagnac interferometers, each of them is sensitive to rotations around one particular axis and a telescope used as highly sensitive star-tracker (10~9rad in the 0.3 to 3 Hz bandwidth). The two units will measure the vector components of the gravitomagnetic rotation rate along the two axes perpendicular to the telescope pointing which is directed to a guide star. The drag variation written above describes the situation for a telescope pointing in the direction perpendicular to the orbital plane of the satellite. The orbit, however, changes its orientation over the course of a year which has to be compensated by a rotation of the satellite to track continuously the guide star. Consequently the pointing of the telescope is not always directed parallel to the normal of the orbital plane. According to the equation, the rotation rate signal will oscillate at twice the frequency of the satellites revolution around the Earth. The modulated signals have the same amplitude (3.75xl0"" 14 rod.s -1 ) on the two axes but are in quadrature. The resolution of the atomic Sagnac units is about 10 _ 1 2 rad.s _ 1 for a drift time of about 3s. Repeating this measurement every 3 seconds the ASU's will reach after 3 hours the level of 10~lirad.s~1, in the course of one year the level of 2.l0~16rad.s~l, i.e. a hundredth of the expected effect.

22

4

Conclusion

Previous experiments measuring the gravitational acceleration of Earth and its gradient or rotations have been demonstrated to be very promising. Sensitivities better than lnrad.s~1.Hz~1^2 for rotation measurements and 2.10~8g.Hz~1/2 for gravity measurement have already been obtained. The sensitivity of matter-wave interferometers for rotations and accelerations increases with the measurement time and can therefore be dramatically enhanced by reducing the atomic velocity. Laser cooling can efficiently reduce the speed of the atoms but cannot circumvent the acceleration due to gravity. On the ground the 1-g gravity level sets clear limitations to the ultimate sensitivities. HYPER-precision atom interferometry in space opens up entirely new possibilities for research in fundamental physics with unprecedented precision. The cold atom interferometers carried by HYPER will be accommodated in a drag-free spacecraft in a low-Earth, Sun-synchronous orbit. The primary scientific objectives of the HYPER mission are to test General Relativity by mapping (latitudinal) structure (magnitude and sign) of the Lense-Thirring effect, to determine the fine structure constant by measuring the ratio of Planck's constant to the atomic mass and to test the equivalence principle on individual atoms, a complement to other space tests of the equivalence principle using massive bodies (STEP, MICROSCOPE). References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.

ESA-SCI(2000)10, July 2000. W. Chow et al, Rev. Mod. Phys. 72, 61 (1985). J.F. Clauser, Physica B 151, 262 (1988). R. Colella et al, Phys. Rev. Lett. 34, 1472 (1975). S.A. Werner et al, Phys. Rev. Lett. 42, 1103 (1979). F. Riehle et al, Phys. Rev. Lett. 67, 177 (1991). M. Kasevich and S. Chu, Appl. Phys. B 54, 321 (1992). Atom Interferometry, ed. Paul R. Berman (Academic Press,1997). T.L. Gustavson, et al, Class. Quantum Grav. 17, 1 (2000). A. Peters, K. Y. Chung and S. Chu, Metrologia 38, 25 (2001). Ch.J. Borde, in Laser sprctroscopy X, ed. M. Ducloy, E. Giacobino, G. Camy, World scientific, 239 (1992). Ch.J. Borde, Phys. Rev. A. 140, 10 (1989). M. Kasevich et al, Phys. Rev. Lett. 66, 2297 (1991). M. Snadden et al, Phys. Rev. Lett. 8 1 , 971 (1998). M. Sagnac, Compt. Rend. Acad. Sci. 157, 708 (1913).

THE GLAST G A M M A RAY LARGE A R E A TELESCOPE S. BREZ, R. BELLAZZINI.N. OMODEI, N. LUMB, L. BALDINI, L. LATRONICO INFN Sez. Pisa Via Livornese 1291 56010 S. Piero a Grado (Pisa) For the Italian GLAST Collaboration

The Gamma-ray Large Area Space Telescope (GLAST) is an international space mission that will study the cosmos in the energy range 10 KeV — 300 GeV, the upper end of which is one of the last poorly observed regions of the celestial electromagnetic spectrum to be explored. GLAST will have an imaging gamma-ray telescope vastly more capable than instruments flown previously. The main instrument, the Large Area Telescope (LAT), will have superior area, angular resolution, field of view, and dead time that together will provide a factor of 60 or more advance in sensitivity, and capability for the study of transient phenomena.

1

Introduction

One of the last bands of the electromagnetic spectrum to be explored for astronomy is the range above 20 MeV. The principal reason for the late start was technological: for energies up to tens of GeV, detectors must be placed in orbit, and even from orbit detection of the low fluxes of celestial gamma rays is difficult. First came EGRET (Launched in 1991): it made the first complete survey of the sky in the 30 MeV — 10 GeV range. The main discoveries of this mission were the detection of gammas with energy > 100 MeV coming from Active Galactic Nuclei (observed more than 60) and the measurement of diffuse gamma ray background to over 10 GeV. But the majority of the sources that shine in the gamma sky don't have a counterpart in low energy: one hundred and seventy sources in the 3rd EGRET catalog are unidentified. GLAST will enable identification of EGRET sources by providing much finer positional error bounds. EGRET raised many interesting issues and questions which can be addressed by a NASA mid-class mission (Delta II rocket). The GLAST mission was conceived to address important outstanding questions in high-energy astrophysics, many of which were raised but not answered by results from EGRET. The main instrument on board the GLAST detector is the Large

23

24

Area Telescope (LAT) that is a pair conversion telescope, like EGRET, but the detectors will be based on solid-state technology, obviating the need for consumables and greatly decreasing instrument deadtime. In this paper we will describe the development of the LAT detector and we will focus our attention in some scientific topics of interest for GLAST. 2

T h e Large Area Telescope

The primary interaction of photons with matter in the GLAST energy range is pair conversion. This process forms the basis for the underlying measurement principle by providing an unique signature for gamma rays, which distinguish them from charged particles. The flux of Cosmic rays, in fact, is as much as 10 times larger. The pair conversion process permits the determination Charged particle anticoincidence shield

Conversion foils

Particle tracking detectors

_=^^

~~1/

-

/

—7 1

/ 1 e+ e-

Calorimeter (energy measurement)

Figure 1. Principle

oj photon detection

in

GLAST.

of the incident photon directions via the reconstruction of the trajectories of the resulting e + e _ pairs. This technique is illustrated in Figure 1 in which the incident radiation first passes through an anticoincidence shield, which is sensitive to charged particles, then through thin layers of high-Z (tungsten) material called conversion foils. The photon converts in these layers producing an electron-positron pair. The trajectories of these charged particles are measured by the tracking detectors, and their energies are then measured by a calorimeter. GLAST was designed to have a low profile to give wide field of view.

25 The Large Area Telescope (LAT) comprises an array of 16 identical "tower" modules (see Figure 2), each with a tracker (Si strips) and a calorimeter (Csl with PIN diode readout) and DAQ module. T h e towers are surrounded by a finely segmented ACD (plastic scintillator with P M T readout) while the support structure is an aluminum strong-back "Grid" with heat pipes for transport of heat to the instrument sides. T h e A n t i c o i n c i d e n c e

Figure 2. The LAT instrument components D e t e c t o r (ACD) has a segmented plastic scintillator to minimize self-veto at high Energy and to enhance the background rejection: the estimated efficiency is greater then 0.9997. T h e purpose of the ACD is to detect incident charged cosmic ray particles t h a t outnumber cosmic g a m m a rays by more t h a n 5 orders of m a g n i t u d e . Signals from the ACD can be used as a trigger veto or can be used later in the d a t a analysis 2 . Each of the 16 T r a c k e r tower modules consists of a stack of 19 "tray" structures. Silicon detector wafers cover either side of a tray with the strips on each side running in the same direction. Every other tray is rotated by 90°, so each W foil is followed immediately by an x, y plane of detectors with a 2 m m gap between x and y layers. T h e detectors are located close to the conversion foils to minimize multiple-scattering errors. T h e b o t t o m tray has a flange to m o u n t on the support grid. T h e electronics hybrids are glued vertically to the tray sides to minimize the gap between towers. Each silicon plane on a tray has a 3 7 c m x 3 7 c m active cross section, giving a total silicon

26

area of 83m 2 (comparable with the ATLAS detector planned for the CERN LHC project). In all there are 11500 Silicon Strip Detectors and a total of 1 million channels. The C a l o r i m e t e r is made of 96CsI crystals (thallium doped) per tower arranged into a hodoscopic, imaging configuration and with PIN diode readout on each end. The electronics chain for each PIN diode is composed of a preamplifier which feeds two shaping amplifiers. Discriminators divide the energy domain into four energy ranges, two peak-detecting track and holds. A third faster shaping amplifier, peaking at 0.5/is is used for fast trigger discrimination. The main features of the calorimeter detector are the large dynamic range (5xl0 5 ), low nonlinearity (less than 2%), low power consumption, and minimal dead time (less than 20/is per event). The LAT trigger is a 3-level system. Primary requirements are high efficiency for all measurable gamma rays, and background reduction to fit with telemetry capacity. Two separate conditions may initiate a hardware trigger for a given tower (LT1). The first request is for the tracker to have three planes hit in a row. The second involves the calorimeter, considering the number of hits in the module. Tower triggers are ORd in the central ACDTEM and fanned out to each tower. The ACD information is optionally used to reduce LT1 rate ("controlled mode"). The second level trigger (LT2) is a tower-based trigger, in parallel for all towers. It uses a fast track finding algorithm and extrapolates track candidates to the ACD tiles to search for vetoes. The veto is not applied to events with large energy deposits in CAL. LT3 is a full instrument event reconstruction trigger. The main features of the three level trigger are summarised in Table 1. Albedo photon events are removed by comparing the reconstructed photon direction with that of the Earth's horizon. The cosmic ray event rate is reduced to less than 15 Hz.

Table 1. T h e 3 different trigger levels adopted in the GLAST Large Area Telescope.

27

3

Summary of GLAST Science Topics

The universe is largely transparent to gamma rays in the energy range of GLAST. Energetic sources near the edge of the visible universe can be detected by the light of their gamma rays. There are good reason to expect that GLAST will see known classes of sources up to redshift 5, or even greater if the sources existed at earlier times. The small interaction cross section for gamma rays can provide a direct view into nature's highest energy accelerators. In addition, gamma rays point back to their sources unlike cosmic rays which are deflected by magnetic fields. The main advantages of the LAT detector will be the wide field of view (2sr) and the extremely short dead time per event (< 100//s). These performances, together with the excellent background rejection (better than 2.5xlO 5 : 1) will allow GLAST to detect both faint sources and transient signals in the gamma-ray sky. The capabilities of the GLAST LAT detector compared to those of EGRET are summarized in table 2. Several performances of the LAT detector, such as the angular and energy resolution, the field of view and the effective area are plotted in figure 3, and compared to those of EGRET. Quantity

EGRET

Energy Range Peak Area Field of View Angular Resolution

20 M e V - 3 0 GeV 1500 cm2 0.5 sr 5.8°

Energy resolution Deadtime per event Source Location Det. Point Source Sensitivity

10% 100 ms 15' lxl0-7cm-2s-1

Table 2. GLAST LAT specification and performance

LAT (Minimum Spec.) 20 MeV-300 GeV 8000 cm2 > 2sr < 3.5° (100 MeV) < 0.15° (> 10 GeV) 10% < 100/is 10 20 eV: (EECR) (Fig.l). A direct question arising is: what is the maximum Cosmic Ray energy, if there is any limit? Addressing the theoretical issue concerning the production and propagation of 1020 eV Primary quanta is problematic and it involves processes still little known. The energy loss mechanism related to the interaction of hadronic particles with the 2.7 Kelvin Universal Radiation Background (Greisen-Zatsepin-Kuzmin effect), conditions the mean free path of Cosmic Radiation. This effect limits the distance of the sources of Primary EECRs to less than 50-100 Mpc, a short distance on a cosmological scale, opening the problems related to the nature of the sources and their distribution in the Universe. Focusing the attention on the primary sources, two general production mechanisms have been proposed for the EECRs: BOTTOM-UP, with acceleration in rapidly evolving processes occurring in Astrophysical Objects. The scenario involves astrophysical objects such as, e.g. AGNs and AGN radio lobes. The study of these objects is, besides radio

38

observations, a main goal of X-ray and Gamma-ray astrophysics of the late 90's. An extreme case in this class is represented by the Gamma Ray Bursts, found to be located at cosmological distances. The observation of "direction of arrival and time" coincidences of GRBs and Extreme Energy Neutrinos (E>1019 eV) in the EUSOmission could provide a crucial test for the identification of the observed GRBs as EECR sources in spite of their location at distances well above the GZK limit. TOP-DOWN Processes. This scenario arises from the cascading of ultrahigh energy particles from the decay of topological defects. Cosmic Strings would play an essential role for releasing the X-bosons emitting the highest energy quarks and leptons. This process could occur in the nearby Universe. The relics of an early inflationary phase in the history of the Universe may survive to the present as a part of dark matter and account for those unidentified EECR sources active within the GZK boundary limit. Their decays can give origin to the highest energy cosmic rays, either by emission of hadrons and photons, as through production of EE neutrinos. From the Astroparticle Physics point of view, the EECRs have energies only a few decades below the Grand Unification Energy (lO^-lO25 eV), although still far from the Plank Mass of 1028 eV. Cosmic Neutrinos, not suffering the GZK effect and being immune from magnetic field deflection or from an appreciable time delay caused by Lorentz factor, are ideal for disentangling source related mechanisms from propagation related effects.The opening of the Neutrino Astronomy channel will allow to probe the extreme boundaries of the Universe. Astronomy at the highest energies must be performed by neutrinos rather than by photons, because the Universe is opaque to photons at these energies. Observational Problems. The extremely low value for the EECR flux, corresponding to about 1 event per km2 and century at E > 10 eV, and the extremely low value for the interaction cross section of neutrinos, make these components difficult to observe if not by using a detector with exceptionally high values for the effective area and target mass. The integrated exposure ( ~ 2xl0 3 km2 yr sr) available today for the ground based arrays operational over the world is sufficient only to show the "ankle" feature at ~5xl0 18 eV in the Cosmic Ray energy spectrum and the existence of about ten events exceeding 10 eV; the limited statistics excludes the possibility of observing significant structures in the energy spectrum at higher energies. Experiments carried out by means of the new generation ground-based observatories, HiRes (fluorescence) and Auger (hybrid), will still be limited by practical difficulties connected to a relatively small collecting area (1019 eV forms a significant streak of fluorescence light over 10-100 km along its passage in the atmosphere, depending on the nature of the Primary, and on the pitch angle with the vertical. Observation of this fluorescence light with a detector at distance from the shower axis is the best way to control the cascade profile of the EAS. When viewed continuously, the object moves on a straight path with the speed of light. The resulting picture of the event seen by the detector looks like a narrow track in which the recorded amount of light is proportional to the shower size at the various penetration depth in the atmosphere. From a Low Earth Orbit (LEO) space platform, the UV fluorescence induced in atmospheric Nitrogen by the incoming radiation can be monitored and studied. Other phenomena such as meteors, space debris, lightning, atmospheric flashes, can also be observed; the luminescence coming from the EAS produced by the Cosmic Ray quanta can be on the other hand disentangled from the general background exploiting its fast timing characteristic feature. EUSO observes at Nadir from an orbital height of about 400 km. It is equipped with a wide angle Fresnel optics telescope (60° full FoV) and the focal plane segmentation corresponding to about 1 km2 pixel size on the Earth surface. The area covered on Earth is of about 160000 km2. Exploiting the high speed of the focal plane detector (10 ns class), EUSO is able to reconstruct the inclination of the shower track by the speed of progression of the projected image on the focal surface and to provide the tri-dimensional reconstruction of the EAS axis with a precision of a degree (or better) depending on the inclination. By measuring the EAS front luminosity with the photoelectrons (PE) detected by the MAPTs covering the focal surface, EUSO registers the longitudinal development of the EAS. EUSO General Requirements and Main Goals For a significant observation from a space mission the assumed values are: a) Effective geometrical exposure of (5> x interaction, and another, by the xdecay, can be seen because of the long enough path-length (~ 1000 [£/1020 eV] km) for x-decays observable by EUSO. Tau-neutrinos above 1015 eV, on the other hand, will be observed and identified as Earth-penetrating "upward" showers (by Cherenkov). High vT flux by the v^ -¥ vT oscillation and the low detection threshold energy for them allow EUSO to make oscillation experiments in space as well as v t astrophysics of AGN above 1015 eV. EUSO Schematic Outline EUSO, originally proposed to ESA for a free-flyer LEO mission , has been approved for an "accommodation study" on the ISS International Space Station. Under the assumption of both a LEO (-500 km altitude) free-flyer mission or the ISS accommodation (400 km average altitude), the coverage of the observable atmosphere surface at the scale of thousand kilometers across and the measurement of very fast and faint phenomena like those EUSO is interested in, requires: - optical system with large collecting area (because of the faint fluorescence signal) and wide equivalent field of view covering a sizable half opening angle around the local Nadir (to reach geometrical factor of the order of 106 km2 sr) , - focal plane detector with high segmentation (single photon counting and high pixelization), high resolving time (-10 ns), contained values for weight and power,

41 - trigger and read-out electronics prompt, simple, efficient, modular, capable to handle hundreds of thousands of channels, and comprehensive of a sophisticated on-board image processor acting as a trigger. Fig.2 shows an artistic view of EUSO attached at Columbus on the ISS

Figure 2. EUSO at the COF-EPF.

/./ EUSO Pavload: The "Main Telescope". The EUSO Main telescope is presented schematically in the artistic view of Fig. 14. The instrument consists of three main parts: Optics, Focal surface detector, Trigger and Electronics System. An effective synergy between the parts constituting the instrument is of fundamental importance for achieving the EUSO scientific objectives. Optics, detector elements, system and trigger electronics have to be matched and interfaced coherently to obtain a correct response from the instrument. Scientific requirements have been of guidance for the conceptual design of the apparatus and in the choice among various possible technical solutions. The design criteria are based on the following assumptions: 380 km orbit FOVof ±30°

Pixel size at ground: 1 km2 Event energy threshold > 5x 1019 eV

42

Figure 3. View of the EUSO Main Telescope.

The observation from space calls for an approach different from that of the conventional ground based fluorescence experiments. For space application the instrument has to be compact as much as possible, highly efficient, and with a builtin modularity in its detection and electronics parts. The Optics. The optical system required for EUSO aims at finding the best compromise in the optical design, taking into account the suitability for space application in terms of weight, dimensions and resistance to the strains in launch and orbital conditions. The optical system views a circle of radius ~220 km on the Earth and resolves 0.8x0.8 km2 ground pixels: this determines the detector size to be adopted to observe the events. The forgiving resolution requirements of EUSO suggest the consideration of unconventional solutions, identified in the Fresnel lens technology. Fresnel lenses provide large-aperture and wide-field with drastically reduced mass and absorption. The use of a broader range of optical materials (including lightweight polymers) is possible for reducing the overall weight. The present Fresnel optical camera configuration study (FoV 60°) considers two plastic Fresnel lenses with diameter 2.5 m and iris diaphragm 2.0 m diameter. The Focal Surface Detector. Due to the large FOV and large collecting area of the optics, the focal surface detector is constituted by several hundreds of thousands of active sensors («2x 105 pixels). The detector requirements of low power consumption, low weight, small dimension, fast response time, high quantum efficiency in UV wavelength (300-400 nm), single photoelectron sensitivity, limit the field of the possible choices to a very few devices. A suitable off-the-shelf device is the Multi-Anode Photomultiplier Harnamatsu R5900 series. These commercial photomultipliers meet closely the requirements imposed by the project. Pixel size, weight, fast time response and single photoelectron resolution are well adaptable to the EUSO focal surface detector. The organization in "macrocells" of the focal

43

surface (a macrocell is a bi-dimensional array of nxn pixels) offers many advantages as easy planning and implementation, flexibility and redundancy. Moreover, modularity is ideal for space application. The Multi-Anode Photomultipliers represent, in this contest, a workable solution. Trigger and Electronics System. Special attention has been given to the trigger scheme where the implementation of hardware/firmware special functions is foreseen. The trigger module has been studied to provide different levels of triggers such that the physics phenomena in terms of fast, normal and slow in time-scale events can be detected. The FIRE (Fluorescence Image Read-out Electronics) system has been designed to obtain an effective reduction of channels and data to read-out, developing a method that reduces the number of the channels without penalizing the performance of the detection system. Expected Results Extensive simulations have been elaborated by O. Catalano at IFCAI/CNR. Fig.4 and Fig.5 report the expected results for EUSO in the ISS version, compared with those referred to the free-flyer version of the original proposal to ESA: in the two versions the results appear almost identical , with the lower altitude for the ISS compensating the reduced dimensions of the optics for what concerns the "threshold".

f/#= 1.15 FREE FLYER; h-500 km tens 0 • 3.5 m

KiKi-gy (eV )

Figure 4 Differential EECR counting rate: comparison between EUSO on the ISS and the original free-flyer proposal. The dashed zone shows the spectral structure induced by the GZK effect.

44

~ 10"

i — i — i — — —

}

1

1—.

•

'

—

—

f/#= 1.15 5 ^ FREE FLYER; h=500 t m ^ O * * ^ tons 0 = S.S m

>

i Jf 1 0 "

'

A\NL

TD V

Greisen v 10' ISS: • 380 km Iens0 » 2.8m

*

10" 10'"

10" En»rny(eV)

Figure 5. Neutrino expectation: the different shadowed areas refer to Topological Defects (TD) v and Greisen v by interaction of the Primary (CR)

A C O M P L E T E SIMULATION OF COSMIC R A Y S A C C E S S TO A SPACE STATION

P. BOBIK, G. BOELLA^M.J. BOSCHIN^, A. FAVALLI, M. GERVASF, D. GRANDI, E. MICELOTTA* P.G. RANCOITA INFN Milano, Italy The Cosmic Rays (hereafter CR) flux we can measure near the Earth is the result of a complex trajectory inside the geomagnetic field. On one side this acts like a shield for low energy primary CR, on the other side middle and high energy CR are focused and driven into preferential directions to reach the Earth. Excluding any kind of energy loss we can say that the energy spectrum in the solar cavity is essentially the same we measure at the Earth, but the magnetic field effect is to transform a (mostly) isotropic flux in a highly asymmetric one. AMS detector on board of the Space Shuttle in June 1998 has observed primary CR mixed with isotropic secondary quasi-trapped in the Earth magnetic field at low (400 km) altitude and over a large Earth surface (80%). We have developed a code to recon. struct the path (both forward and back in time) of CR inside the magnetosphere (hereafter mags). We realized a complete simulation of the primary CR flux seen by AMS in 1998. In this way we will obtain a relation between the input directions of primary CR outside the mags, and the observed (AMS data) directions. This transfer function F (R, d, ip) is related to the rigidity R of the particle, and to the detecting position.

1

Introduction

1.1

Cosmic Ray Access

CR detectors design and performance are different in relation to their location. If the detector is located on the Earth surface his aim to study primary CR flux and composition for example will be persued only with the detection of secondary CR products of the shower occurred in the atmosphere. This will be followed by the reconstruction of the primary vertex obtaining the energy of the particle from the number of secondaries. On the contrary CR space detectors are not affected by the interaction of particles with the atmosphere and are able (at least in principle) to detect the primary CR component without a reconstruction of the secondary CR chain. In between lie balloon borne experiments that are similar to space detectors strategy but unable to measure over a wide Earth surface because of the balloon lifetime. In •UNIVERSITY OF MILANO BICOCCA, ITALY tCILEA SEGRATE, ITALY tUNIVERSITA DEGLI STUDI DI MILANO, ITALY

45

46

addition the balloon is not in orbit, like a satellite but it floats at the top of the atmosphere. These primary CR detectors (see Sec. 1.2) measuring CR flux and composition obtain the following information: mass and charge of the particle, kinetic energy (or equivalent rigidity 0 ) and arrival directions. All these quantities are related to a well determined detection position and attitude (very important especially for space detectors that cover a wide part of the Earth's surface). These primary CR follow a very complex path from their source to their detection. CR are modulated when enter the heliosphere in relation to their energy and charge (the effect is larger at low energies). When CR approach the Earth they are focused by the geomagnetic field as a function of the rigidity when enter the mags. In the present contribution we are interested in the latter effect: the trajectory of primary CR in the mags from the magnetopause 6 (hereafter magp) to a space detector. We implemented a code x that reproduces the trajectory of charged particles in the Earth mags in order to study the CR flux behaviour (rigidity spectrum as a function of arrival directions) at the magp. 1.2

AMS detectors

Our aim was to use the tracing code in order to simulate CR access to a Space detector like AMS-01 was (and AMS-02 will be). We then used AMS-01 CR fluxes 2 to renormalize the flat spectrum used as input in our simulation. AMS-01 offers a unique opportunity to realize our simulation in a complete way. In fact AMS-01 data has a high statistics, wide Earth surface covered both in latitude and longitude and wide rigidity range. AMS detector flow in space on board of the Space Shuttle in June 1998 in the STS-91 mission. In 10 days of data taking it collected over 108 CR protons in addition to electrons, positrons and Helium nuclei3. We then started reproducing the primary CR flux at the AMS altitude to obtain a relation with their original entering points at the magp. Moreover the new AMS-02 detector is in preparation. This detector will be equipped with a superconductive magnet, an electromagnetic calorimeter, a TRD detector and a RICH Cherenkov counter in addition to a TOF and a Tracker. This completely redesigned detector will be detecting CR for three years on board of the ISS Space Station starting from 2004-2005. The new sensitivity of AMS will allow the separation of isotopes probably up to the "substantially it is the particle momentum divided by his charge ''the border of the magp, a region of reconnection between the solar magnetic field carried by the solar wind and Earth magnetic field

47

Beriullium. This new mission will have different geomagnetic and solar characteristics and this will be studied in a predictive simulation that will start in the next months. In addition the distribution of primary CR flux on the magp will enable us to trace forward these particles and let them interact with the Earth atmosphere in order to reproduce the non negligible secondary CR flux seen by AMS-01. 2 2.1

The Simulation Particle tracing

The tracing code is based on the solution of the Lorentz equation of a charged particle in a static magnetic field starting from particle velocity:

FLorentz=m—

= Z-q-y$

xBtotal\

(1)

In effect geomagnetic field is not static, but eq. (1) can be anyway applied in our case with the requirement of a small time step (typically -^) in relation to the instant gyroradius of the particle (see Subsec. 2.2 for errors). In this way the magnetic field can be considered locally constant and eq.(l) solved. In this code updated internal (IGRF 2005) and external (Tsyganenko 96) magnetic field models are used, together with a magp analytical shape (Sibeck '91) and a Runge-Kutta 6th order metod to solve Lorentz differential equation. Tsyganenko 96 4 external field model is based on all existing data and reproduces all currents of particle present inside the mags (Birkeland regions as well as tail field and ring current). The Sibeck shape of the magp is an ellipsoid toward the Sun and a cylinder opposite to it. We also introduced the Tsyganenko correcting factor that enables this fixed surface to vary in function of the solar wind pressure (than confines the geomagnetic field). IGRF internal field model is a NASA standard world wide used, based on all geomagnetic measurements and updated every five years. Our main pourpose was to have primary CR information (rigidity and incoming directions) at AMS altitude. In order to prevent long time useless calculations we added the code a subrotuine to compute for every particle his Stoermer 5 rigidity cut-off:

i W 7 , A ) = 59.6-

1 — \f1 — cos 7 • cos A3 cos 7 • cos A

(2)

48

where Rcut is in GVolts, and the angles are: 7 the incident particle East-West angle, A the geomagnetic latitude 6 . Then if particle rigidity was below 70% of rigidity cut-off we jumped trajectory calculation and considered the particle as secondary CR. 2.2

Simulation

Structure

The main structure of our simulation was due to three different goals: • the results have to be easily renormalized • the simulation should be the most complete as possible • the calculations should be fast for a future on-line application To fulfill these request we elaborated a structure as follows. Positive particles (protons, the main CR component) are generated at AMS altitude. They are back-traced in time until they reach one of the two boundaries: the magp or the same generation sphere. In the first case particles are primary CR, otherwise secondaries. To shoot particles uniformly over the sphere we divide it in 3600 parts, with a spacing that vary in latitude with cos# in order to have the same elementary shooting surface (or ess). Prom each of these ess we chosed 1800 shooting angles equally spaced in a hemisphere (same method as for ess in order to have the same solid angle). Finally rigidity (or kinetic energy) bins for protons were chosen exactly equal to AMS ones: 31 log bin from 0.3 GV (0.07 GeV) to 200 GV (199.6 GeV). With this choices we were able to minimize the intrinsic loss of information hidden in such a work. To have primary CR flux incident on AMS detector in fact the first idea was shoot particles from the border of the magp. The efficiency in this case is very low (~ 10~ 3 — 10 - 4 ) only few particles reach the AMS sphere. A similar approach could be shooting particles closer to the Earth (2RE where RE is the Earth radius) but again the efficiency for primary CR is not high (50%) and a renormalization with data is needed. Our efficiency is almost (100%) because particles are generated at the same place and same conditions of renormalization data. No method can prevent from uncertainty. Here we summarize our errors due to the code and the simulation structure: • the choice to jump trajectory calculation for particles with low rigidity (see Sec. 2.1) introduced a systematic error in primary CR identification of 10- 2 • the internal algorithm error is ~ 10~ 4 and is evaluated with the calculation of the difference between final point rigidity and start rigidity. • finally we checked for angular resolution for primary CR, tracing a fixed amount of particles up to the magp and then reversing the velocity and tracing

49

them back to the Earth, this gives an accuracy of ±0.5° We will build the so-called transfer function that depends by 10 variables, of which six are related to the detection point and four to the magp entering point. F{R,rdet,'ddet,iPdet,6inc^inc,'&mag

,ent)- This will be a

10 i/l dimension matrix in principle of 2 • 108 elements, many of which will be empty. This simulation was realized using two Alpha stations (DS10 and DS20), a Linux farm (5 Linux with 1,2 Ghz AMD Duron CPU) and distributed CPU (INFN Condor facility). 3

Results

We renormalized our simulation following the geomagnetic coordinates separation as in AMS results 7 . We selected only data within a cone of 32° for zenith facing to compare with data 7 , and then we reconstructed a primary and secondary CR spectrum from our flat spectrum of generated events (see Fig. 1). In our simulation an isotropic flux of CR's on the detector has ben asumed. To a first approximation this hypothesis is reasonable at least for secondaries, while primaries could have a smooth east-west modulation. We find that, mostly in low geomagnetic latitudes, secondary CR flux is present up to several GeV (~ 20) in downgoing protons (see Fig. 2) and is smoothly decreasing with increasing geomagnetic latitude (see Fig. 2 right). Actually the primary CR flux has a cut-off steeper than what appears from the overall measured flux. In addition a high energy peak appears for the secondary CR flux. Another result of our simulation is the distribution of magp origin points as a function of particle rigidity (see Fig. 3 and Fig. 4). There for instance we can see that at rigidity lower than 15 GV origin points on the magp are concentrated close to the shifted-tilted geomagnetic equator 6 , while for higher rigidities (see Fig. 3) the distribution is much more isotropic. References 1. 2. 3. 4. 5. 6. 7.

P. Bobik et al, Proceedings of Vulcano Workshop 2000 461, 387 (2001). The AMS Collaboration, Phys. Lett. B 461, 387 (1999). The AMS Collaboration, Phys. Lett. B 494, 10 (2000). N. A. Tsyganenko, JGR 100, 5599 (1995). K. Stoermer, Z. Astroph. 1, 237 (1930). http://hpamsmi2.mib.infn.it/ wwwams/geo.html. The AMS Collaboration, Phys. Lett. B 472, 215 (2000).

50

AMS CR proton flux

I

• IOJ £ 0.2 rod • 0,2 5 I0J £ 0.3 rod

1

10''

4 1


ux oscillations, with parameters A m 2 = 2 x 1 0 - 3 eV 2 and s i n 2 ( 2 0 ) = 1.0. From left to right: L/E spectra for upward muon events (hatched area) and downward ones (open area); their ratio with the best-fit superimposed and the result of the fit with the corresponding allowed regions for oscillation parameters at 68%, 90% and 99% C.L..

have comparable values and the L/E resolution is intrinsically limited by the knowledge of the incoming neutrino path (L). Oscillations of muon neutrinos should manifest themselves in a modulation of the L/E spectrum (see Figure 1) from which the oscillation parameters can be measured. The experimental technique 4 described above has sensitivity to J/M oscillations with Am 2 > 6 x 10~ 5 eV2 and mixing near to maximal, fully covering the region of oscillation parameters suggested by Super-Kamiokande results. 3

D e t e c t o r and E x p e r i m e n t a l a p p r o a c h

To ensure the observation of a sufficient number of events, the detector should have at least an overall mass of 30 kt. The proposed detector layout consists in 125 horizontal iron layers (8 cm thick) interleaved with active detector elements housed in 2.2 cm gaps. The total detector dimension will be around 15 x 13 x 30 m 3 depending on the final assembly location. The detector is designed for high energy event containment (up to tens of GeV). To include in the analysis also the outgoing events (Partially Contained) a strong magnetic

71

Figure 2. Left: Expected allowed regions of u^ — vr oscillation parameters for MONOLITH after four years of exposure, The results of the simulation for A m 2 = 0.7,2,5,8,30 x 1 0 - 3 eV 2 and maximal mixing are shown. Right: MONOLITH exclusion curves at 90% and 99% C.L. after one or 4 years of data taking assuming no oscillations. The full (dashed) black line shows the published results of the Super-Kamiokande (Kamiokande) experiment.

field (1.3 T) has been included in the detector design. The neutrino energy is obtained measuring the corresponding muon momentum by range (Fully Contained events; Ap^/p^ « 8%) or by the track curvature (Partially Contained events;ApM/pj, « 20%) with the requested accuracy. The direction of the incoming neutrino (up or down) and the related vertex can be determined through tracking with fast timing. The muon charge measurement can be used to study potential matter effects differently affecting neutrinos and antineutrinos. The design of magnetic field has been studied in order to preserve detector modularity and active elements accessibility. Glass RPC 5 counters, will be used as active detector elements. Every plane will provide two coordinates through crossed 3 cm strips. A time resolution of ss 1 ns and a linear behaviour of the reconstructed energy up to 10 GeV has been measured 6 ' 7 .

72

4

Glass R P C

The Glass RPC present design5 consists in a pair of commercial float glass electrodes 243 mm wide, 1.85 mm thick and up to 2 meters long. The volume resistivity p at room temperature of this glass is about 1012 ficm, suitable for operation in streamer mode in low particle rate environments. Injection molded spacers that are used to ensure the 2 mm distance between the electrodes. Each spacer has 2 sticks respectively 2 mm and 150 mm long, that are orthogonal to the 200 mm long spacer profile. The spacers are clamped to the glass with the sticks inserted between the electrodes. The shape of the stick is knurled to prevent possible discharge between the plates. Spacers are placed every 100 mm in such a way that they channel the gas flow throughout the active volume. A water based graphite varnish is applied to the external surfaces of the glass to provide the high voltage supply. The surface resistivity of the graphite coating is about 400 kfi/square. The glass plates and the spacers are inserted in an extruded envelope with 1.5 mm thick walls. The external cross section of the envelope is 250 x 9 mm2. The detector is closed by 2 injection molded end caps. The H.V. supply to each electrode is ensured by a properly shaped harmonic metal strip located in one of the two end caps. With respect to the bakelite RPC, this detector is more suitable for large production because of the reduction in necessary manpower. In fact, the float glass electrode does not need the surface treatment with linseed oil; the spacers are applied without gluing; the H.V. contacts are realized without soldering. The use of an envelope for the gas containment instead of a glued frame between the electrodes reduces the occurence of leakage. In order to obtain in a large experiment a good time resolution, the electric field between the electrodes must be uniform over the whole sensitive surface of the experiment. The signals from the Glass RPC are generated faster as the electric field increases with a slope of At(ns)/AE(V/mm) w 10~ 2 . This means that a time resolution of the order of 1 ns is possible only if the electric field is uniform at the percent level; consequently, a gap tolerance less than few tens of microns is necessary. The sticks of the spacers have a mechanical tolerance of ± bum. The effect of the gravity, of the electrostatic attraction, and of the gas pressure has been taken into account to achieve an adequate gap precision. As the maximal distance between the sticks of the spacers is 100 mm, the glass sagitta due to gravity in the detector is less than 1.5 /xm. The sagitta due to the pressure of the flowing gas is absent in the Glass RPC because the gas can at most inflate the envelope, without any deformation (or damaging) of

73

the electrodes. The sagitta due to the electrostatic attraction, assuming an electric field of 4 kV/mm, is less than 5 jum for each glass. The gap tolerance in the Glass RPC is hence less than 0.5% and does not strongly depend on the construction procedure or on the working conditions. The readout is based on the new idea of using fiat cables as pick-up strips. The pick-up strip width of 3 cm is determined by the front end electronics, where 24 flat cable conductors are short circuited together and sent to each channel.

5

Conclusions

After experiment approval, the first module can be operative and acquire data within four years. The detector completetion will occur in another two year while the first module is continuously recording events. After one year of data acquisition with the detector completed we expect to obtain first indications to confirm or exclude the Super-Kamiokande region (see Figure 2). Finally, four years of data taking will allow to either completely exclude the region proposed by Super-Kamiokande or to deduce the oscillation parameter Am 2 for large mixing angles. As shown in Figure 2, the progress in the measure of Am 2 is comparable to the one obtained by Super-Kamiokande with respect to Kamiokande. This achievement qualifies MONOLITH as a next generation neutrino detector. Glass RPC detectors have been developed and engineered. They are suitable for mass production and match the requirements for large neutrino experiments. References 1. MONOLITH Proposal, LNGS P26/2000, CERN SPSC 2000-031, SPSC/M657. 2. Super-Kamiokande Collaboration, Y. Fukuda et al., Phys. Rev. Lett. 81 (1998) 1562. 3. M. Aglietta et al., LNGS-LOI 15/98, CERN/SPSC 98-28, SPSC/M615, Oct. 1998. 4. P. Picchi and F. Pietropaolo, "Atmospheric Neutrino Oscillations Experiments", ICGF RAP. INT. 344/1997, Torino 1997, (CERN preprint SCAN-9710037). 5. C. Gustavino et al, Nucl. Instr. and Method A 457 (2001) 558. 6. M. Ambrosio et al., Nucl. Instr and Method A 456 (2000) 67. 7. G. Bencivenni et al., Nucl. Instr and Method A 461 (2001) 319.

DEVELOPMENT AND PERFORMANCES OF T H E M A G I C TELESCOPE D . Bastieri* C. Bigongiari, F . Dazzi, M. M a r i o t t i , A. Moralejo* L. P e r u z z o , A. Saggion a n d N . Tonello Dipartimento di Fisica - Universita di Padova & I.N.F.N. - Sezione di Padova, Via Marzolo, 8 - 35131 Padova - Italy The MAGIC Collaboration is building an imaging Cerenkov telescope at La Palma site (2200 m a.s.l.), in the Canary Islands, to observe gamma rays in the hundredGeV region. The MAGIC telescope, with its reflecting parabolic dish, 17 m in diameter, and a two-level pattern trigger designed to cope with severe trigger rates, is the Cerenkov telescope with the lowest envisaged energy threshold. Due to its lightweight alto-azimuthal mounting, MAGIC can be repositioned in less than 30 seconds, becoming the only detector, with an adequate effective area, capable to observe GRB phenomena above 30 GeV. MAGIC telescope is characterised by a 30 GeV energy threshold and a sensitivity of 6 x l 0 - 1 1 c m - 2 s - 1 for a 5c-detection in 50-hours of observation. In this report, some future scientific goals for MAGIC will be highlighted and the technical development for the main elements of the telescope will be detailed. Special emphasis will be given to the construction of the individual metallic mirrors which form the reflecting surface and the development of the fast pattern-recognition trigger.

Introduction Several energy decades of the electromagnetic spectrum are available for observation by modern astronomers: from radio waves to gamma rays of roughly 100 TeV there is virtually no interruption, if it weren't for a gap between ~ 10 GeV, the upper energy limit of satellite-borne detectors, and ~ 300 GeV, the energy threshold of ground-based Cerenkov telescopes. Whereas increasing the upper energy limit of satellite-detectors clashes with the poor statistics of observable gamma-rays, reducing the energy threshold of Cerenkov detectors is, in principle, only a matter of available technology, as Cerenkov detectors, based on calorimetric observation of cosmic rays that interact in the atmosphere, rely on huge effective factor 104 bigger than their satellite counterparts. Strong scientific motivations support the idea that a lot of interesting physics lies in this gap (see Saggion and Bastieri, 2001), forcing the physics community to take the burden of exploring new technologies to lower the energy threshold of present Cerenkov detectors. * Corresponding author: denis.bastieri®pd.infn.it. * Work supported by the Spanish MECD.

74

75

Two of the main ingredients to obtain a so-called second-generation Cerenkov detector, that is a single Cerenkov telescope or an array of Cerenkov telescopes with an energy threshold well below 100 GeV, are the reflecting surface and the trigger architecture. While a huge reflecting surface is necessary, as the energy threshold scales down with an increase of area, a fast and clever trigger is desirable, in order to sift incoming events and relax data acquisition requirements. One of the four second-generation Cerenkov detectors planned to be working in the next few years is MAGIC, the Major Atmospheric Gamma Imaging Cherenkov Telescope, that, with its 17 m 0 of reflecting surface is the biggest, and with the lowest energy threshold, single telescope. German, Italian and Spanish institutions make up most of the MAGIC Collaboration and are presently building the telescope which should see the first light in the Summer 2002 (seeLorenz, 2001). 1

Future scientific goals

An energy threshold as low as the MAGIC one, allows a thorough study of AGNs up to z « 3. There are currently two different pictures of AGNs: the satellite one, with EGRET identifying dozens of AGNs at different redshifts, and the ground-based one, with only two safely-detected AGNs at about 250 GeV and z « 0.03. This could be a hint of AGN power-law spectra cutting off between 20 and 250 GeV, or it can be entirely ascribed to the interaction of emitted 7-rays with the extragalactic photon background in the IR/optical/UV bands. A thorough AGN study can thus shed light on AGN cut-offs and, as well, on the photon background. Moreover, beyond presently well-known AGNs, a systematic study on gamma emitters may be widened to comprehend other unidentified EGRET sources, as well as supernova remnants (SNRs), pulsars (PSRs), microquasars and X-ray binaries. This can help in identifying the main sources of cosmic rays up to an energy of 106 GeV. It has to be mentioned also the fact that Cerenkov experiments deeply rely on Montecarlo, as far as energy reconstruction is concerned: an eventual overlap of the observable energy range with satellite experiments, that detect directly the primary cosmic rays, can be thus of great help. GRBs are another possible issue for MAGIC: in fact, the lightweight design of the supporting cradle of MAGIC permits a repositioning of the whole telescope in less than 30 seconds. This gives MAGIC the unique opportunity of an early follow-up of GRBs at energies in the ten GeV window (see Bastieri et al., 2001).

76 Radii of curvature of MAGIC mirrors .

~|

| integ = 97I

.

. ^^^

•

AM thp

'Ili-irs

M l r r r o e s on Juno

Figure 1. Curvature radii of all the individual mirrors of the MAGIC telescope. In evidence the ones already produced and tested (updated 22-Nov-2001).

MAGIC can also explore the realm of particle physics; in fact, the probability of exotic interactions should clot around the galactic centre, denser, and may allow the identification of new particles such as WIMPs or the lightest SUSY particles, via the observation of annihilation lines (in the gamma region) toward the galactic centre. 2

The reflecting surface

The reflecting surface of MAGIC is a tessellated F = 1 parabolic dish of 17 m 0 . It is made up with 972 spherical mirrors with the curvature radius that best matches the paraboloid locally (see figure 1). Each mirror measures 50 x 50 cm2 of area and is made according with aeronautical techniques: an aluminium honeycomb fills an outer box, also in aluminium, made with a plied plate 1 mm thick and 2.5 cm high. The box supports a "thick" aluminium plate (5 mm), later to be worked into the desired spherical shape. Between this plate and the honeycomb, a defrost printed circuit is inserted and everything, the

77

plate, the defrosting circuit, the honeycomb and the box is assembled in a high-pressure tank using a laminating adhesive to glue parts together. The upper plate is machined exploiting the so-called fly-cutter technique. The panel is fixed on a rotating table, while the cutting tip spins off-axis on a tilted T-shaped tool. The curvature radius of the worked plate goes as: "T-tool" radius . . ., . rr• sin(tilting angle) The machining (milling) is made in various steps of increasing refinement, with the roughness becoming ~ 10 nm after the last step. After being milled, the panel is tested to see whether it meets the production requests and eventually sent to be quartz coated against accidental scratches, dust and ageing. curvature radius =

3

The trigger

Trigger requirements can be understood by analysing the event taxonomy. Events can be grouped into three main classes: Night Sky Background (NSB), shower events coming from a hadron progenitor and events coming from electromagnetic showers, the only class of interest to MAGIC. NSB events are characterised by having only few, scattered photoelectrons and by not having any next-neighbour logic appearance. On the other hand, shower events have many photoelectrons, but while hadronic events are quite lumpy in appearance, e.m. events are more uniform and quite definitely of elliptic shape. The trigger groups together 271 pixels in the inner part of the camera. Electronic constraints suggest a partition of the pixels into 19 macrocells of 36 pixels each (a 4-pixel side hexagon without an outer pixel) with some overlap among the pixels (see figure 2). The trigger is arranged into two levels and the macrocell unit is common to both levels for an easier signal cascading. The first level makes a tight time coincidence (2-^5 ns) on simple patterns of close packed images to rejects NSB. The hearth of the first level is made with 4 Programmable Logic Devices (PLDs) per macrocell, each of which takes care of recognising in its area a 2, 3, 4 or 5 fold next-neighbour logic. The first level enables also the second level. The second level performs some clever pattern recognition eventually enabling the data acquisition chain. The total transition time from first level to the beginning of the DAQ chain is less than 100 ns. Level 2 is arranged in a cascading structure of macrocells, where each macrocell contains a Look Up Table (LUT) that codes, according to user requests, the input pattern. In this way it is possible to mask unwanted region of the camera (bright stars), to perform a minimal image analysis (centre of gravity and rough pixel/cluster

78

271 inner pixels inside the trigger 19 macrocells of 36 pixels each

180 2" PMTs 397+18 1" PMTs

Figure 2. The trigger architecture. The image sketches the top-down structure of the trigger, from the 415 + 180 camera pixels to the overlapped covering of the inner camera with 19 macrocells, to a single macrocell with 36 inputs.

counting) useful for a fast hadron rejection or even to give focus only to a determinate camera region, desirable, for instance, for pulsar studies. Acknowledgements The authors wish to thank all the people of the firm LT Ultra of Aftholderberg, Germany, for their patience and collaboration during the milling and the onsite optical-quality measurement of the individual mirrors. References 1. A. Saggion and D. Bastieri, "The Observational Energy Gap between 10 and 300 GeV", to appear in Procs. Vulcano 2001, Italy, 2001. 2. E. Lorenz for the MAGIC Coll., 27 th ICRC, Hamburg, 2001, 2789-2792. 3. D. Bastieri et al, 27 th ICRC, Hamburg, 2001, 2759-2762. 4. D. Bastieri et al, Nucl. Instrum. Methods A 461, 521 (2001).

N E W D E V E L O P M E N T S OF P H O T O D E T E C T O R S FOR T H E LAKE BAIKAL N E U T R I N O E X P E R I M E N T

B. K. LUBSANDORZHIEV Institute for Nuclear Research of Russian Academy of Sciences, pr-t 60-letiya Oktyabrya 1A, Moscow 117312, Russia E-mail: [email protected] New developments of photodetectors for the lake Baikal neutrino experiment are described. Some test results of photodetectors at the lake Baikal are presented.

1

Introduction

The lake Baikal Neutrino Experiment has history of more than 20 years, starting from small short experiments with a few PMTs in the early 80s to the present large scale longterm operating neutrino telescope NT-2001, which has been put into operation on April 6th 1998. The telescope's effective area for muons is 2000-10000 depending on a muon energy. The rate of events due to atmospheric neutrinos is about 1 per two days.

2

Neutrino telescope NT-200

The lake Baikal Neutrino Telescope NT-200 is located in the southern part of the great Siberian lake Baikal at 3.6km from the shore and at the depth of 1km. The schematic view of NT-200 and optical modules attached at a string are presented in fig.l. The telescope consists of 192 optical modules at 8 vertical strings arranged at an umbrella like frame. Optical modules are grouped in pairs and swithched in coincidence with 15ns time window. So two optical modules in one pair define one optical channel resulting in rather low optical channel background counting rate of 100-300Hz. Two pairs of optical modules form so called svyazka. The detector electronics system is hierarchical: from the lowest level to the highest one - optical module electronics, svyazka electronics module, string and detector electronics modules. In the latter detector trigger signals are formed and all information from string electronics modules are received and sent to the shore station. Three underwater electrical cables connect the detector with the shore station. The detector is operated from the shore station.

79

80 To Shore

— calibration laser • ~~ array electronics "" module j, -,— string F «L electronics * T module •f-^-OMs tp # X ' ^

svjaska electronics module

Figure 1: The lake Baikal neutrino telescope NT-200

3

Quasar-370 p h o t o t u b e

The Quasar-370 phototube 2>3>4>5 is a hybrid phototube which consists of an electro-optical preamplifier with larg hemispherical photocathode and small conventional type PMT, see fig.2. Photoelectrons from a large 37cm diameter hemispherical photocathode with 2TT acceptance are accelerated by 25 kV to a fast, high gain luminescent screen (YSO scintillator is usually used). The light flashes in luminescent screen induced by photoelectrons are read out by small PMT with 3cm diameter photocathode. The latter has been developed especially for this kind of application by INR and MELZ factory 6 . As a result one photoelectron from the hemispherical photocathode yields typically 25 photoelectrons in the small PMT.This high gain first stage results in an excellent single photoelectron resolution. Due to the fast acceleration of primary photoelectrons by 25 kV high voltage and mushroom shaped glass envelope the time jitter can be kept rather low. Last but not the least the tube is almost insensitive to the Earth's magnetic field. Averaged over more than 200 tubes the mean values for single photoelectron resolution and time

81

Figure 2: Quasar-370 phototube

resolution is 70% and 2ns respectively. Quasar-370 phototube is characterized by conspicuously low level of afterpulses in comparison with conventional PMTs. The level of afterpulses in Quasar-370 tube is substantially less than 1%. Another advantage of Quasar-370 phototube in contrary to conventional PMTs is the lack of prepulses due to the fact that the first stage of the phototube is optically separated from tube's photocathode. It was shown 2 ' 3 , 5 that Quasar-370 parameters depend strongly on characteristics of a scintillator in a luminescent screen. So recently we have developed a number of modifications of Quasar-370 tube with new scintillators. The most promising results we have got with SCBO3, YAP and LSO. Unfortunately the latter one has one substantial drawback for using in Quasar-370 tube. It has rather high value of Zeff. It is important to note here we need scintillators with Z e / / as low as possible to suppress effectively late pulses r due to photoelectron backscattering effect. With these new scintillators we have reached Ins (PWHM) time jitter and 40% single photoelectron resolution. It is noteworthy to mention here about very much intriguing scintillatoi8 - ZnO : Ga with less than Ins decay time and light yield of 40% of Nal :Tl. At present we try to work with this scintillator and to reproduce results of8. The success of this work would undoubtedly be a kind of a breakthrough in designing of very fast and effective hybrid phototubes.

82

DISK:LIM_back/A0O15.MON

N

1

20 18 16 14 12 10 8 6 4 2 0

20 18 ftjpm***^***^^*****

iJ 16

o 14 12 10 8

6 4 2 i

•

•

•

20.6

•

i

•

20.7

20 18 16 14 12 10 8 6 4 2 0

•

•

•

Started: 0 6 / 0 4 / 0 1 20130

i

20.8

*»*tm'**i*jM*rtJiAJi*it»**tfl

20.6

20.7

20.8 time, hours

• * * *

1

•

20.6

•

•

•

I

•

20.7 20.8 M u o n T r i g g e r Rate vs t i m e , C h a n n e l

Figure 3: Muon Trigger rate versus time. Left up and left bottom - for conventional Baikal optical channel. Right up - for new optical channel based on new two-channel optical module.

4

Two-channel optical module

The pairwise ideology pursued in NT-200 neutrino telescope has many advantages. The ideology allows to suppress effectively an individual optical module background counting rate due to water luminescence and a phototube's dark current, to eliminate phenomena deteriorating phototube's time resolution namely prepulses, late pulses and afterpulses. Moreover such an approach facilitates very much designing of trigger system, data acquisition system etc. Unfortunately there is just one but essential shortcoming. Namely it's too expensive to have two optical modules for one optical channel. To overcome this problem we have developed two-channel optical module based on two-channel version of Quasar-370 phototube 9 . Two-channel Quasar-370 phototube uses the same electro-optical preamplifier and new two-anode small PMT with mesh dynode system instead of conventional PMT. This new two-anode small PMT has following characteris-

83

tics: time jitter - a = 340ps; Peak-to valley ratio of single photoelectron sharge distribution - 1.5; cross-talks between channels - 1%. Basing on two-channel Quasar-370 phototube we have developed two-channel optical module which incorporates besides Quasar-370 phototube two fast transimpedance preamplifiers for each anode signals and uses the same glass pressure vessel and the same penetrators as conventional Baikal optical module. In the course of the last expedition at the lake Baikal we have tested one pilot sample of two-channel optical module in frame of a new experimental string designed to test technological innovations for future neutrino telescopes at the lake. In the test measurements at the lake we used ordinary but slightly modified NT-200 front-end electronics. Output signals of each channel are switched in coincidence just in the same way as in NT-200. It results in the same local trigger counting rate of 100-300Hz. Fig.3 presents the time dependence of muon trigger rate for two conventional Baikal optical channels and new optical channel based on two-channel optical module. One can see that the new optical channel has comparable sensitivity to muons with ordinary optical channels of NT-200. It opens new possibilities for NT-200 further extension plans. 5

Other developments

A new version of Quasar-370 phototube (Quasar-370D) with semiconductor diode as a photoelectron multiplying element instead of a system of luminescent screen and small PMT has been developed. We have manufactured two pilot samples which are under studies now. Modifications of Quasar-370 phototubes (Quasar-370G)10 which are able to withstand high current due to night sky background operate successfully in the wide angle atmospheric air Cherenkov detectors TUNKA11 and SMECA12. Quasar-370L is low background version of Quasar-370 phototube. It has U238 and Th232 content of about 10~ 8 g/g. Quasar-370L is aimed at using in low background experiments. Another development is concerned with phototubes with high quantum efficiency (more than 40% in the range 450-550nm) photocathodes and it's results will be reported elsewhere. Conclusion The lake Baikal Neutrino Experiment successful operation proves high performance and high reliability of a number of photodetectors developed for this kind of application. New developments are focused at future neutrino experiments at the lake Baikal and other experiments in high energy physics, cosmic ray and astroparticle physics.

84

Acknowledgments This work was supported by the Russian Ministry of Research (contract 102ll(OO)-p), the German Ministry of Education and Research and the Russian Fund of Basic Research (grants 99-02-1837a,01-02-31013 and 00-15-96794) and by the Russian Federal Program "Integration"(project 346). The author is indebted very much to his colleagues from KATOD and MELZ laboratoties and Baikal Collaboration. The author would like to thank particularly Dr. V.Ch.Lubsandorzhieva for many invaluable advices and help in preparation of this paper. References 1. I.A.Belolaptikov et al, Astropart. Phys. 7 263 1997 2. L.B.Bezrukov et al Proc. 3rd NESTOR Workshop, Pylos 1993. Univ.Athens. 132 1994 3. R.I.Bagduev et al Proc. Int.Conf. "Trends in Astroparticle Physics", Teubner, Stutgart, 132 1994 4. R.I.Bagduev et al, Nucl. Instrum. Methods A 420, 138 (1999) 5. B.K.Lubsandorzhiev, Nucl. Instrum. Methods A 442, 368 (2000) 6. L.B.Bezrukov et al, Instrum. and Experim. Techn. 1 104 2000 7. B.K.Lubsandorzhiev et al, Nucl. Instrum. Methods A 442, 452 (2000) 8. PHILIPS Photomultiplier tubes, 6-35 1994 9. B.K.Lubsandorzhiev et al, Proc. of the 25th ICRC.7 269 1997 10. B.K.Lubsandorzhiev et al, Instrum. and Experim. Techn. 3 104 2001 11. N.Budnev et al, Proc. of the 25th ICRC.2 581 2001 12. V.A.Balkanov et al, Yadernaya Fizika 63 1027 2000

OPERATION A N D CALIBRATION OF LARGE-MASS D R O P L E T D E T E C T O R S FOR PICASSO M. BARNABE-HEIDER, N. BOUKHIRA, M. DI MARCO, P. DOANE, M-H. GENEST, R. GORNEA, C. LEROY, L. LESSARD, J.-P. MARTIN, H. PAQUETTE, V. ZACEK Groupe de Physique des Particules, Departement de Physique, Universite de Montreal, C.P. 6128, Succ. "Centre-Ville", Montreal (Quebec) H3C 3J7 Canada; Corresponding author: [email protected] The PICASSO cold dark matter (CDM) detector is based on the phase transition produced by nuclear recoils in room temperature superheated liquids, induced by CDM particles, such as neutralinos predicted by supersymmetric models. Largemass superheated droplet detectors have been built for the first time. We review their properties and operation. Simulations performed to understand the detector response are presented, briefly. Signal definition and analysis are described together with possible sources of acoustic background, with their frequency signatures, and eventual elimination. The PICASSO innovative droplet detectors for CDM search will allow the quantitative study of one of the main questions of modern physics.

1

Introduction

Information from the anisotropy measurements of cosmic background radiation and the galactic recession velocities as measured with large red-shift supernovae suggests that Cold Dark Matter (CDM) consists mainly of nonbaryonic particles *. Weakly interacting massive particles (WIMP) are ideal candidates for CDM. Among them stands the neutralino, predicted by minimal supersymmetric models. The neutralino is stable if R-parity is conserved and has a mass of the order of 100 GeV/c 2 . A mass lower limit of 50 GeV/c 2 has been extracted from LEP experiments 2 . In their gravitational motion around the center of the Galaxy, the neutralino velocity distribution is Maxwellian with v r m s ~ 300 km/s. At the position of the solar system, they are supposed to be the main ingredient of the measured average mass density of 0.3 GeV/cm 3 . They interact very weakly with nuclei, which then recoil with typical energies within a range 3 of 0 - 100 keV. The concept of the PICASSO experiment 4 rests on an approach to detect CDM candidate particles, the neutralino in particular, through the use of large mass superheated droplet modular detectors. The operation of these detectors, similar to that of bubble chambers, is well described by the theory of Seitz 5 . They are threshold detectors, and their sensitivity to various types of radiation is strongly dependent on operation temperature and pres-

85

86

sure. Their high efficiency for detecting neutrons, for an adequate gas at room temperature, features a detection process based on the energy deposited by the nuclear recoil produced in the collision of a heavy particle, the neutralino ultimately, with a nucleus of the active medium. This nuclear recoil triggers the liquid-to-vapour phase transition of room-temperature superheated carbo-fluorates, while the droplet detectors are relatively insensitive to minimum ionizing particles and to nearly all sources of background 6 . The very low interaction cross sections between CDM and the detector active medium nuclei requires the use of very massive detectors to achieve a sensitivity level allowing the detection of CDM particles in the galactic environment. This article presents a review of the features of the large mass superheated droplet detectors.

2

The superheated droplet detectors and their operation

The large-mass droplet detectors consist in an emulsion of room temperature meta-stable superheated freon-like (C4F8, C4F 10 , etc) droplets dispersed in an aqueous solution. This solution is subsequently polymerized after dissolution of an appropriate concentration of a heavy salt (e.g., CsCl, NaBr, sodium acetate) in water. The salt is used to equalize densities of droplets and solution. Bubbles are stationary after formation due to the gel elasticity and can be recompressed to liquid droplets for a new round of measurements by applying pressure with a piston or with compressed gas to the detector container. Bubble formation is triggered by the heat spike deposited when a particle traverses a length of superheated liquid. This formation can be measured either visually (when the loading or detector volume are small enough) or acoustically via piezo-sensors sensitive to the pressure wave produced during the explosive phase transition. A detector is shown in Fig. 1 in a container capable of holding pressures up to 10 bars. Piezo-sensors are glued on the container surface for signal detection. Typical gas loading is in the 10-40 g/litre range. The detectors have been built, according to PICASSO specifications, by BTI (Bubble Technology Industries) 7 . Superheated droplet detectors are threshold detectors. Their response is determined by gas thermodynamic properties and depends on operating temperature and pressure. The detector operation can be understood in the framework of the theory of Seitz 5 which poses that bubble formation is triggered by the heat spike resulting from the energy deposition when a charged particle traverses the superheated medium. The potentiel barrier, E c , which prevents spontaneous liquid-to-gas transition in a superheated liquid, is given

87

J % Figure 1. A 1.5 litre detector module equipped with sensors.

by the Gibbs equation

Ec

. SE-fflL

(1)

3 {Pi - p0)z where po, pj (functions of temperature, T) are the applied and internal pressures, respectively. The surface tension is given by a(T) =OQ ( T C - T)/(T C To) where T e is the critical temperature of the gas, J ... "•*'•

T

;0^

-

' 1"' i

i

•t-i M W

! 1-— j

^

.

— .,_ Figure 5. Results of vibration tests on a SuperGlast tray. LEFT: resonance search results; RIGHT: notched random vibration acceptance spectrum

performed at the end of the test showed no variation in the resonance peak and a visual inspection of the tray confirmed that it successfully survived to the dynamic test without any damage. 4

Conclusions

GLAST prototype trays of the SuperGlast type were subjected to both thermal and dynamic tests to verify their ability to survive the mission environment. The preliminary test results have shown a successful behaviour of the tray structure, that will be confirmed by further studies in the near future. References 1. General Environmental Verification Specification for STS & ELV payloads, subsystems and components, NASA Goddard Space Flight Center; http://essp.gsfc.nasa.gov/essplib/pdf/gevs.pdf 2. F. Giordano, M.N. Mazziotta, S. Raino , LAT TKR Tray Test Plan, TKR Group Internal Note LAT-TD-00154 (unpublished); http://www-glast.slac.stanford.edu/documents/LATDocSort.asp 3. S.Ney, E.Swensen, E.Ponslet, Hytec Technical Note HTN-102050-018, 8/3/2000 (unpublished)

This page is intentionally left blank

101

Silicon Tracker Organizers: W. Braunschweig K. Freudenreich P. Weilhammer K. Freudenreich A. Bean E. Kajfasz F. Palmonari I. Fiori S. D'Auria C. Cecchi W. Wallraff A. Lusiani C. Krauss A. Polini S. Cabrera L. Moroni Yu. Gornushkin P. Netchaeva D. Ferrere M. Biasini M. Lenzi A. Dierlamm V. Zhukov T. Rohe N. Zorzi E. A. F. B.

Berdermann N. Georgobiani Simon Grube

(Tracker Ill-Parallel) (Tracker II-Parallel) (Tracker I-Parallel)

Convener's Report Design of an Upgraded D 0 Silicon Microstrip Tracker for Fermilab Run2B DO Silicon Microstrip Tracker for Run IIA The CDFII Silicon Tracking System The CDF Online Silicon Vertex Tracker Commissioning and Operation of the CDF Silicon Detector The Assembly of the AMS Silicon Tracker, Version 1 and 2 The AMS Infrared Tracker Alignment System - from STS91 to ISS Performance of the BaBar Silicon Vertex Tracker Charged Particle Tracking with the HERA-B Detector The ZEUS Micro Vertex Detector The Run IIB Upgrade of the CDF Silicon Detectors The BTeV Pixel Detector System Monolithic CMOS Pixels for Charged Particle Tracking Status and New Layout of the ATLAS Pixel Detector The ATLAS Silicon Microstrip Tracker Construction Status The Silicon Strip Tracker of the CMS Experiment The CMS Silicon Tracker Automated Module Assembly CMS Silicon Tracker - Milestone 200 Test of the CMS Silicon Strip Detectors in the Hadron Beam Status of the CMS Pixel Detector Fabrication of Microstrip Detectors and Integrated Electronics on High Resistivity Silicon The Diamond Project at GSI - Perspectives Radical Beam Gettering Epitaxy of ZnO and GaN GEM Detectors for COMPASS Architecture of the Common Gem and Silicon Readout for the COMPASS Experiment

This page is intentionally left blank

A R E V I E W OF SILICON D E T E C T O R S (Convener's Report)

K. FREUDENREICH Laboratory for High Energy Physics, ETH Zurich, Switzerland The status of silicon strip detectors which were presented at the 7th International Conference on Advanced Technology and Particle Physics is reviewed.

1

Present silicon detectors

Due to their superior performance in b-tagging, silicon detectors are widely used by accelerator-based high energy experiments, at least for the innermost tracking. At this conference, reports from experiments running at e + — e - - , at e-p-, at proton-, p-p- and at heavy-ion-machines were presented in the parallel sessions. In addition, future experiments at LHC, in space and at a future linear collider were discussed as well. The reports from the running experiments were particularly interesting, since their experience may guide future developments. Usually the inner detectors are arranged in concentric rings around the beam line. Fig. l(left) shows the inner silicon tracker from CDF l for run IIA. In order to reduce the background from synchrotron radiation in the horizontal plane the two HERA detectors HI and Zeus 2 have their detectors arranged asymmetrically around the beam pipe (see Fig. 1 (right)).

Figure 1. Transverse view of the barrel detector of the CDF experiment (left) and of the barrel detector of the Zeus experiment (right).

103

104

In contrast to these, the Phobos silicon detector (Fig. 2 left) 3 at RHIC looks quite different. It consists of a cylindrical detector around the beam pipe and two spectrometer arms. Fig. 2 (right) shows the illumination of the detectors from gold on gold collisions, the interaction point is clearly visible. Every 1 - 2 hours some strings of chips have to be switched off when the detector currents get too high.

"sir

*a

Figure 2. Schematic view of the PHOBOS silicon detector (left). Illumination by a gold on gold collision (right).

Presently, most experiments use double-sided silicon detectors. Due to their more complex structure, double-sided detectors are limited in the bias voltage they can sustain. Therefore, one can not fully compensate eventual radiation damage by increasing the bias voltage. Both experiments \ 4 at the Fermilab Tevatron expect significant radiation damage of their silicon detectors after 4 fb" 1 . Therefore, CDF 5 and DO 6 will replace their present detectors (only the 7 inner layers in the case of CDF) with single-sided ones for ran IIB. Fig. 3 shows the layout of both Fermilab experiments for that run. The new detectors for run IIB will be actively cooled (-10° C in the case of DO). High radiation damage was reported for the detectors even at e + - e~ storage rings. Detectors in layer one of the CLEO III detector 7 developed unexpected radiation damage. It is not yet clear whether this is due to non-standard wafers or not. Detectors in the innermost, horizontal plane of the BaBar silicon detector 8 also were reported to be at the limit of their radiation

105

damage budget in the beginning of the run.

Figure 3. Transverse view of the new barrel detectors for run IIB. DO (left) and CDF(right)

In order to profit from the high resolution of the silicon detectors, the position of these detectors has to be known very precisely. There were two talks on the laser alignment of silicon detectors. The AMS experiment 9 uses special alignment sensors which are coated anti-reflectively and which have narrow, 10 jj, wide readout strips to keep the transparency as high as possible. In the AMS-G1 detector, it was possible to record a laser ray in 6 successive, 300 /J, thick layers. A similar system is planned to be installed in the CMS experiment 10 . A newly installed trigger system n in the CDF detector allows the tagging of events with secondary vertices at the trigger level. It reduces the level 1 trigger rate of 5 MHz to 50 Hz at the level 3 trigger. An online impact parameter resolution of 69 /u is achieved (see Fig. 4).

108 1200

1000

-

39.43 1 12 Coss'ar.' 1039. ± Mess jj 1) I20JB-021 Si&ma &_ O0.J1UB±

a

800

600

-

J

400

J

\

200

0

14.90 QJQ79E-84

,«.«•»»*«••««« Aiwnf -0.1 -0.075 -0.05 -0.025

0

0.025

0.05

0.075

0.1

Figure 4. Distribution of the online impact parameter in the CDF experiment.

In general, the newly installed detectors at the various accelerators perform as expected. Their results can be found in the respective contributions to this conference. Fortunately, also problems with the installation and running in were reported 4 , 1 2 . For example, in the CDF experiment at Fermilab, some cooling lines in the outermost layer (ISL) of the CDF detector were reported to be blocked *. With the help of a laser some glue could be removed and CDF hopes to recuperate all cooling in subsequent shut-downs. From the experience of running detectors the following recommendations can be made: 1. Simple modular design. 2. Ability to stand high voltage. The first two requirements lead both to the choice of single sided detectors. 3. Observation of strict quality control at each step of the production of the detector. 4. Minimizing the handling and transportation of sealed detectors. 5. Cooling of the detectors to improve their radiation hardness. 6. Use of rad-hard read-out chips. Chips in 0.25 /x technology have improved radiation hardness and are also cheaper. The detectors now being built for the Fermilab run IIB and for LHC will follow these basic rules.

107

2

Future silicon detectors

The size of the silicon detectors has and will increase significantly with respect to the former LEP detectors as can be seen from Table 1. experiment CMS ATLAS CDF II DO 2 ZEUS LHCb VELO AMS2 Delphi Babar Phobos Aleph L3 Belle SVD

silicon area [m2] 210 61 5.8 -> 8.7 4.7 -» 8.3 0.32 6.5 1.8 0.95 0.49 0.23

nb. of strip readout channels xlO 6 9.6 6.3 0.722 - • 1.083 0.793 - • 0.952 0.207 0.205 0.196 0.175 0.140 0.137 0.095 0.086 0.082

Table 1. Comparison of the surface covered by silicon detectors and the number of strip readout channels for some experiments. The two numbers for CDF and DO refer to run IIA and run IIB respectively.

The two major detectors at the LHC, ATLAS 13 and CMS 14 , will have very large silicon detectors. In order to cope with the big number of modules, CMS has developped a system which will assemble and glue the components of a module by a robotic machine 15 . Both experiments will have pixel detectors in the innermost layers for precise vertex reconstruction followed by strip detectors which will allow to obtain high momentum resolution in the bending plane. In the case of CMS the tracker will entirely consist of silicon detectors. Both experiments will use single sided detectors. Stereo detectors will be made out of two single sided detectors put together at an angle. In order to keep the signal to noise ratio constant, the long strip length modules of CMS in the outer part will consist of 500 fi thick detectors instead of the usual 300 /i thick ones. The detectors of both experiments will be cooled down to -7° C (ATLAS) and -10° C (CMS). Fig. 5 (left) shows a perspective view of the 5.6 m long and 1 m diameter strip detector for the ATLAS experiment. Fig. 5 (right) shows the side view of one quarter of the 5.6 m long and 2.2 m

108

diameter strip detector of the CMS experiment. Both experiments will cover a pseudo-rapidity range of ± 2.5.

^m I * ? :£l sfV £•&•"

Figure 5. Perspective view of t h e silicon detector for the ATLAS experiment at LHC (left) as seen from one endcap. Side view of one quarter of t h e CMS experiment at LHC (right). Dimensions are indicated in mm. T h e pseudo-rapidity is indicated as well.

Finally, first results from a monolithic active pixel (MAP) detector were presented 16 . The MAP silicon sensor combines charged particle detection and readout electronics on the same substrate. First tests with minimum ionizing particles showed high signal-to-noise ratios and excellent position resolution (see Fig. 6).

,((im)

Figure 6. Difference between t h e position measured by a beam telescope and by a M A P detector.

109 3

Summary

Within a decade silicon detectors have increased both in surface and in the number of readout channels by at least one order of magnitude. At the LHC another increase by one order of magnitude will occur. Today's baseline design of silicon detectors comprises single-sided detectors which are actively cooled and read-out by rad-hard electronics in sub-micron technology. Silicon detectors are now being used in high level b-tagging triggers. Monolithic pixel detectors have promising prospects. References 1. F. Palmonari, The CDF II Tracking System, these proceedings. 2. A. Polini, The Zeus Micro-Vertex Detector, these proceedings. 3. G. J. van Nieuwenhuizen, Performance results of the PHOBOS silicon detectors, talk in the parallel session. 4. E. Kajfasz, DO Silicon Microstrip Tracker for Run IIA, these proceedings. 5. S. Cabrera, The Run IIB Upgrade of the CDF Silicon Detectors, these proceedings. 6. A. Bean, Design of an Upgraded DO Silicon Microstrip Detector for Fermilab Run IIB, these proceedings. 7. H. Kagan, New Results on the Radiation hardness of CLEO Silicon, talk in the parallel session. 8. V. Re et al., Performance of the BaBar Silicon Vertex Detector, these proceedings. 9. W. Wallraff, The AMS Infrared Tracker Alignment System - from STS91 to ISS, these proceedings. 10. W. Braunschweig, Alignment System of the CMS Endcap Structure, talk in the parallel session. 11. A. Bardi et al., The CDF Online Silicon vertex Tracker, these proceedings. 12. S. D'Auria, Commissioning and Operation of the CDF Silicon Detector, these proceedings. 13. D. Ferrere, The ATLAS Silicon Microstrip Tracker Construction Status, these proceedings. 14. M. Biasini, The Silicon Strip Tracker of the CMS Experiment, these proceedings. 15. M. Lenzi, The CMS Silicon Tracker Automated Module Assembly, these proceedings. 16. Yu. Gornushkin, Monolithic CMOS Pixels for Charged Particle Tracking, these proceedings.

D E S I G N OF A N U P G R A D E D D 0 SILICON M I C R O S T R I P T R A C K E R FOR F E R M I L A B R U N 2 B A. BEAN FOR THE D 0 COLLABORATION Department of Physics and Astronomy, University of Kansas, Lawrence, KS 66045, USA; E-mail: [email protected] The D0 collaboration has begun designing a new large scale silicon strip tracking system for the D0 detector for Run2b of the Fermilab Tevatron collider, which will deliver an integrated luminosity in excess of 15 fb _ 1 to the experiment. The current design of the replacement detector will be presented with emphasis on those issues that drive the design. The tracker will employ about 2200 single sided silicon sensors. Detector readout will utilize the SVX4 chip that is currently being designed for the CDF and D0 Run2b upgrades.

The Higgs boson plays a unique role within the Standard Model that describes the known properties of elementary particles. It alone is responsible for breaking the gauge symmetry and generating masses for the gauge bosons and fermions. A growing body of evidence suggests that the Higgs boson is relatively light, and may well be within the reach of the CDF and D 0 experiments at the Fermilab Tevatron Collider. A crucial piece of instrumentation for the Higgs search is a silicon vertex detector which can efficiently identify secondary vertices associated with the b-quarks from a light Higgs decay. The D 0 collaboration has designed a new silicon vertex detector for use in the so-called Fermilab Run2b era that is optimized for the Higgs search. The design of the Run2b silicon detector is based on an optimization of the physics performance of the detector while at the same time satisfying various boundary conditions, both external and internal. Using the Tevatron parameters for Run2b, the inner layer is chosen to have a fiducial length of 96 cm which results in an acceptance of more than 96% of the delivered luminosity. The design operating temperature of the inner layer was chosen to be -10 degrees Celsius based on irradiation studies and the behavior of the leakage current, depletion voltage, and equivalent noise charge. Our operating voltages are expected to be well below 1000V. We also determined that a minimum radius of about 18 mm for the innermost layer of silicon will allow for an adequate safety margin for running the detector to integrated luminosities of 15 fb _ 1 . The Run2a silicon detector employs a silicon track trigger that processes data from the Level 1 central track trigger and the silicon tracker. The trigger observes a 6-fold ^symmetry which must be preserved in the upgraded design. The total number of readout modules in the new system is

110

111

constrained by the currently available cable plant, which allows for about 940 readouts. The proposed silicon detector has a 6 layer geometry arranged in a barrel design. The detector will be built in two independent half-modules joined at z=0. The six layers, number 0 through 5, are divided in two radial groups. The inner group, consisting of layers 0 and 1, will have axial readout only. Driven by the stringent constraints on cooling, these layers will be grouped into one mechanical unit called the inner barrel. These layers have a significantly reduced radius relative to the current tracker. Given the tight space constraints, emphasis has been placed on improving the impact parameter resolution. The outer group is comprised of layers 2 through 5. Each outer layer will have axial and stereo readout. The outer layers are also important for providing stand-alone silicon tracking with acceptable momentum resolution in the region 1.7 < \rj\ < 2.0 where D 0 has good muon and electron coverage but lacks coverage in the fiber tracker. The outer layers are assembled in a mechanical unit called the outer barrel. The inner barrel is inserted into the outer barrel forming a half-module. A half-module is the basic unit that is installed in the collision hall. While all 6 layers are designed to withstand 15 f b - 1 of integrated luminosity with adequate margin, separating the inner layers into a separate radial group provides a path for possible replacement of these layers. The outer layers should easily withstand luminosities up to approximately 25 fb _ 1 . Of paramount importance to the successful construction of the new detector in the less than 3 years available, is a simple modular design with a minimum number of part types. This is one of the reasons that single-sided silicon sensors are used throughout the detector. Only three types of sensors are foreseen. All of the sensors are envisioned to have axial traces with intermediate strips. The stereo readout in the outer layers will be accomplished by tilting the sensor slightly with respect to the beam axis. Figure 1 shows an axial view of the Run2b silicon tracker. The inner two layers have 12-fold crenellated geometry and will be mounted on a carbon fiber lined, carbon foam support structure. The sensors in layer 0 will be two-chip wide (128 channels per chip), 78.4mm long with 50 /xm readout pitch. Because of the lack of space available and the cooling requirements for the innermost layer, no readout electronics will be mounted on the sensors. Analog cables will be wirebonded to the sensors carrying the analog signals to a hybrid where the signals will be digitized and sent to the data acquisition system. Keeping the hybrid mass out of the detector active region also helps in reducing photon conversions. The depletion voltages at the end of the run are expected to be around 600V for the innermost layer.

112

Figure 1. Axial view of the proposed tracker upgrade. The outer four layers provide both axial and stereo track measurements, while the inner two layers provide only axial measurements.

The bias system for the inner layers will be designed to deliver voltages up to 1000V. Layer 1 will contain 3-cMp wide sensors that are 78.4mm long with 58 (an readout pitch. The geometry matches the segmentation of layer 0. Although the heat load from putting hybrids directly on the sensors is greatly increased, noise and production considerations have led to on-board electronics for all layers except layer 0. In Layers 2-5 only one type of sensor will be used. The sensors will be 5 chips wide, 100mm long with 60 fim readout pitch. This pitch allows for direct bonding between SVX4 chips and the sensors. Retaining the i n e resolution in Layer 5 significantly improves pattern recognition. These layers employ stiff stave support structures. A stave will have carbon fiber sheets mounted on an inner core that will carry the cooling lines. Silicon will be mounted on the carbon fiber sheets; on one side there will be axial readout and on the other small-angle stereo.

113

Figure 2. Plan view of the Ruu2b silicon tracker inside of the central fiber tracker (CFT).

The longitudinal sementation is driven by the need to match JJ coverage throughout the detector up to r) = 2. For Layer 1 six sensors form one half length in z matching the coverage for LO which is in the same mechanical structure. For Layers 2 and 3 five sensors will form a stave. Staves consist of six sensors for Layers 4 and 5. Figure 2 shows a plan view of the tracker inside the fiber tracker. A decision was made by both CDF and D0 to read out the new silicon systems using the SVX4 chip. This chip is based on the SVX3 chip, but will be produced in 0.25 micron technology. This chip is intrinsically radiation hard and is expected to be able to withstand the radiation doses incurred in the innermost layers. In order not to have to redesign the entire D0 data acquisition and trigger system, the SVX4 chip will be read out as the SVX2 chip is presently. The SVX2 chip is the readout chip for the Run2a detector and incurs deadtime on every readout cycle unlike the SVX3 chip that can run in a deadtimeless mode. By studying the deadtime it was determined that for Layers 2 and higher we are able to read out a total of 10 chips. This allows us to use a double ended hybrid design that reduced the hybrid readout count

114 Table 1. Design parameters of the Run2b Silicon detector. There are a total of 2184 sensors and 888 hybrids in this design. Layers 2-5 have both stereo and axial readouts. The labels A and B refer to sublayers within each layer.

R(mm) Layer OA OB 1A IB 2A 2B 3A 3B 4A 4B 5A 5B TOTAL

Nphi 12 12 12 12 12 12 18 18 24 24 30 30

18.6 24.8 34.8 39.0 53.2 68.9 86.2 100.3 116.9 130.6 147.0 160.5

# Sens. in z 12 12 12 12 10 10 10 10 12 12 12 12

# Sens. Total 72 72 72 72 120 120 180 180 288 288 360 360

Sens. Width(mm) 15.50 15.50 24.97 24.97 41.10 41.10 41.10 41.10 41.10 41.10 41.10 41.10 2184

#Read. in z 12 12 12 12 8 8 8 8 8 8 8 8

Total Chips 144 144 216 216 480 480 720 720 960 960 1200 1200 7440

#Hyb. Total 72 72 36 36 48 48 72 72 96 96 120 120 888

by a factor of two. We plan to use ceramic hybrids using beryllia. The digital signals will be launched onto a jumper cable from the hybrid to a "junction" card located at the end of the active region and then read out through twisted pair cables to an "adapter" card located on the cryostat face. The design parameters are summarized in Table 1. In layers 2-5 there will be a total of 168 staves, containing 336 readout modules. For comparison, the Run2a silicon detector has 793,000 readout channels while the Run2b one will have 952,000 readout channels. Comparisons between the Run2a and Run2b detectors show that the major difference is found at the inner and outer radii. By decreasing the radius of the innermost layer from 25.7mm to 18.6mm, the impact parameter resolution is cut by a factor of 1.5 for Run2b. Because we are removing a lot of the cable plant from the Run2a barrel modules, we are able to utilize this space at larger radii for silicon sensors. The increase from 94.3mm to 163.6 mm for the outer radius allows us to put in two more layers of tracking necessary for the pattern recognition in the Run2b environment. We have modelled the detector with a full GEANT simulation and fully reconstructed events with the expected tracking algorithms. With the new detector we will have better stand-alone silicon tracking with the capability of discovering the Higgs up to a mass of 185 GeV/c 2 .

DO SILICON MICROSTRIP T R A C K E R FOR R U N IIA E. KAJFASZ - FOR THE DO COLLABORATION FERMILAB, MS 310, P.O. 500 Batavia, IL 60510, USA CPPM, Case 907, 163 Avenue de Luminy, 13009 Marseille Cedex 9, FRANCE E-mail: [email protected] We discribe the production, installation and commissioning of the new 792,576 channel DO Silicon Microstrip Tracker to be used for the 2 f b - 1 of the Run Ha at the Tevatron.

1

Introduction

DO has built a Silicon Microstrip Tracker (SMT) to help reach its physics goals for Run Ha of the Tevatron during which, in the next 2 to 3 years, it is supposed to collect 2 f b - 1 worth of data. Construction of the SMT was finished in December 2000 and installation was completed for the beginning of Run Ha in March 2001. The following sections will discuss the different phases of the project. 2

Design

One main improvement included in the DO detector upgrade 1 for Run Ha is its central tracking system as shown in Fig. 1. It includes a 2 T superconducting solenoid, a Central scintillating Fiber Tracker (CFT) and the SMT. The SMT design is driven by two classes of events. Barrels and central disks cover the ~ 25 cm RMS long luminous region for high pr central physics (\r)\ < 1.5). Forward disks are implemented mainly to study b-physics in the forward region down to pseudo-rapidities" of 3. The SMT is comprised of 6 barrels, each barrel mated on one of its ends to an F-disk, 2 stacks of 3 F-disks (end disks modules) and 4 H-disks (see Fig. 1). The barrels are 12 cm long and have 72 ladders arranged in 4 layers (12,12,24,24), each layer having 2 staggered and overlaping sub-layers (see Fig. 2). The 2 outer barrels have single sided (SS) and double sided 2° stereo (DS) ladders. The 4 inner barrels have double sided double metal (DSDM) 90° stereo and double sided "Pseudo-rapidity is defined as rj = — ln[tan(|)] where 6 is the angle w.r.t. the beam direction.

115

116

Co'orinPter ECU

F orward Prp-Shoner

CFPS)

Inter Cryostat D e t e c i o r

Figure 1. DO Central Tracking System.

2° stereo ladders. The ladders are mounted and aligned to 10 — 20/*m between 2 precision machined Be bulkheads. The bulkhead supporting the side of the ladders carrying the read out electronics is equipped with cooling channels. The F-disks are made of 12 wedges of double sided stereo detectors. The H-disks are made of 24 pairs of single sided detectors glued back to back. For the disks, the wedges are mounted and aligned on Be rings which include cooling channels. The barrels and F-disks are precisely mounted in 2 carbon fiber cylinders which meet at the nominal interaction point in the DO detector. The 4 H-disks are individually mounted in carbon fiber cylinders. Tables 1 and 2 summarize some SMT design numbers. Assemblies made of Kapton

Table 1. SMT numbers (module means ladder or wedge). Channels Modules Si area Inner radius Outer radius

Barrels 387,072 432 1.3ro2 2.7em 9.4cm

F-disks 258,048 144 0.4m 2 2.6cm 10.5cm

H-disks 147,458 96 pairs 1.3m2 9.5cm 26cm

117

Table 2. SMT detector types (module means ladder or wedge). Location Barrel layers: L1,L3 (outer bar.) L1,L3 (inner bar.) L2,L4 Disk wedges: F H

Module type

Stereo angle (°)

Pitch

#of modules

#chips /mod

#of HDIs

SS DSDM DS

0 0/90 0/±2

50 50/150 50/60

72 144 216

3 3/3=6 5/4=9

72 144 216

DS SS

+15/-15 +7.5/-7.5

50/60 50/50

144 96

8/6 6/6

288 192

Figure 2. SMT barrel geometry.

flex circuits laminated to Be substrates (High Density Interconnects or HDIs) are used to hold the SVXIIe 1.2/xm rad-hard technology read out chips. The SVXIIe has 128 channels, each with a 32 cell analog pipeline and an 8-bit ADC. It features 53 MHz read out speed, sparsification, downloadable ADC ramp, pedestal, and bandwidth setting 2 . 3

Production, assembly, and testing

HDI flex circuits are electrically tested, laminated to Be substrates, stuffed with component and SVXIIe chips. The stuffed HDIs are electrically tested for functionality and performance (pedestal, noise, gain of every channel, sparsification...) and burned in for 2 to 3 days. In parallel, sensors are tested (CV curves, leakage currents, bias resistors ...) and selected using probe stations. To build a ladder, we use a construction fixture to glue an HDI to silicon sensors. The gluing process is performed on a CMM to align the sensors within a few microns to the edges of the mounting notches which reference

118

the ladder w.r.t. the barrel bulkheads. Once the ladders or wedges are glued, their sensors are wire bonded to their SVXIIe read out chips. Altogether, the number of wire bonds in SMT amounts to more than 1.5 million. The ladders/wedges are then electrically tested, repaired if necessary, burned in and laser scanned. The laser scan allows to measure their operating voltage and identify their dead channels. Averaged over all ladders/wedges, SMT has less than 2% dead channels. Selected ladders or wedges are then mounted onto barrel Be bulkheads or disk Be rings with a position accuracy of about 20/j.m. The production started in May 1999 and ended in October 2000. It was mainly paced by HDIs fabrication and stuffing problems on one hand, and silicon sensor yields, delivery delays and fabrication problems (e.g. sensor p-stop isolation lithography defects for the DSDM sensors, or p-side microdischarges due to misalignemnt of the Al strips w.r.t. the p + implants, worst in the case of DS 2°-stereo sensors) on the other hand. The barrels and disks assembly and their installation in their respective carbon fiber cylinders were completed by December 2000. 4

Readout

Fig. 3 shows how the read out of the SMT is set up. The HDIs are connected through 2.5m long Kapton flex cables to Adaptor Cards located on the face of the Central Calorimeter. The ACs transfer the signals and power supplies of HDIs to 10m long high mass cables which connect to Interface Boards. The IBs supply and monitor power to the SVXII chips, distribute bias voltage to the sensors and refresh data and control signals traveling between the HDIs and the Sequencers. The Sequencers control the operation of the chips and convert their data into optical signals carried over l G b / s optical links to VME Readout Buffers boards. The VRBs receive and hold the data pending a Level-2 trigger decision. The maximum L2-accept rate at DO will be lKHz, corresponding to a data output rate of ~50Mb/s. 5

Installation a n d commissioning

Barrels and F-disks were installed in the DO detector by December 2000. The last H-disk was installed early February 2001. The final cabling was completed in April 2001. Initially, 15% of the 912 HDIs could not be read out. However, during the October 2001 Tevatron shutdown, we managed to repair most of them. 95% are now fully functional. The cooling system was gradually lowered to its nominal temperature to study possible adverse effects on CFT light yields. The cooling system uses a mixture of 30%-glycol/70%-

119

Figure 3. SMT read out.

water circulated at a nominal -10°C, so the detectors run between - 5 ° C and 0°C when powered. We optimized the timing and SVXII chip download parameters to maximize the signal to noise ratio. Calibration procedures and programs (pedestal, noise, gain, sparsification threshold) have been implemented. Data was successfully taken with magnet on or off, with the part of the CFT which was instrumented, and with all the other detector subsystems. Track matching between SMT and CFT shows that the tracker are inter-aligned within 40/UTO, 6

Conclusions

The DO SMT was assembled and installed on time for the start of run Ha. We used the time until October 2001 to commission and understand the detector hardware and its online/offline software. Now we are ready to make full usage of it and enjoy the physics goals it will allow us to reach. By the end of run Ila, after 2fb - 1 , because of radiation damage, the first layers of the SMT will not be of much use anymore. DO is already working on a replacement Silicon Microstrip Tracker for Run lib the design of wich should allow it to accomodate an integrated luminosity in excess of 15fb - 1 3 . References 1. The DO Upgrade: The Detector and Its Physics, Fermilab Pub-96-357-E. 2. T. Zimmerman et al., The SVXII Readout chip, IEEE Trans. Nucl. Sci. N S 42 (1995) 803. 3. A. Bean, these proceedings.

T H E CDFII SILICON T R A C K I N G S Y S T E M F.PALMONARI ON B E H A L F O F T H E CDFII* COLLABORATION INFN

St.Piero

a Grado, via Livornese vecchia 1291, 56100, Pisa, E-mail: [email protected]

Italy

The CDFII silicon tracking system, S V X , for Run II of the Fermilab Tevatron has up to 8 cylindrical layers with average radii spanning from ~(1.5 to 28.7) cm, and lengths ranging from ~(90 to 200) cm for a total active-area of ~ 6 m 2 and ~ 7.2 x 10 5 readout channels. SVX will improve the CDFII acceptance and efficiency for both B and high-Pt physics dependent upon b-tagging. Along with the description of the SVX we report some alignment survey data from the SVX assembly phase and the actual status of the alignment as it results from the offline data analysis. The problems encountered are also reviewed.

1

Introduction

The SVX project developed between 1992 and 2001; final assembly started in 1999; on January 16 2001, SVX was installed into CDFII and run Ha started in early march 2001 a . SVX is made of 3 mechanically separate subdetectors LOO, SVXII and ISL that exploit different features of the same radiation-hard chip SVX3D for data acquisition 1 ' 2 (DAQ). The original design of SVXII (4 layers of double sided sensors) was modified to add a fifth layer and the Intermediate Silicon Layers (ISL) and LOO have been added later on (1996 and 1998).

2

The SVX tracker

In Figure 1 (right) we show the SVX location in the CDFII detector. The concept of the SVX tracker is that LOO and SVXII provide vertex information at radial coordinates below ~10 cm; LOO inside SVXII (mounted on the beryllium beam-pipe) can enhance charged track's impact parameter (do) resolution; SVXII and ISL provide three-dimensional hits (r,tp,z); the SVXII ones can feed real-time informations to SVT 3 for trigger purpose while the *FOR CDFII AUTHORS LIST: HTTP://WWW-CDF.FNAL.GOV/CDFAUTHORS.HTML "Tevatron collider is designed to provide collisions at y/s = 2.0 TeV center of mass energy with maximum istantaneous luminosity of 2 x 10 3 2 cm~2s~1 and interbunch spacing that nowadays is 396 ns and will be reduced to 132 ns.

120

121

forward6 (backward) ISL are used to track particles where the outer tracking (COT) coverage is not complete. In this way the SVX improves and extends the CDPII tracking capabilities outside COT M l coverage limits,

Figure 1. (left) Geometry layout of the SVX=LOO+SVXII+ISL silicon layers; (right) Location of SVX inside CDFII for the Run Ila of Tevatron (Fermilab)

Efforts where made to minimize the amount of material. In LOO, Figure 2 (left), lightweight fine-pitch kapton cables (50 /im kapton film, 10 jum trace width and maximum length of 47 cm) allow all of the front-end electronics (hybrids) to be moved outside the tracking volume. The LOO support structure is made out of a multi-layer layup of high thermal conductivity carbon fiber (CF) with integrated cooling0 for the silicon sensors (wafers). The SVXII is arranged into three identical barrels symmetrically mounted with respect to the interaction point. Within each barrel, Figure 2 (center), the low mass constraint is addressed: gluing hybrids on the wafers, integrating cooling channels into the ladder's d beryllium supports (bulkheads) and arranging comunication electronics (portcards) in a external ring. For ISL, Figure 2 (right), the development of two kinds of CF supports (one for each layer) allowed the building of ~50 cm long modules (i.e. selected4 pairs of ladders joined together) with hybrids at opposite ends. This minimized the transversed material in the central region. In addition, the ~2 m long for ~7.3 kg CF space frame5 (SF, the supporting structure of ISL), provides the three mount points for the SVXII CF-honeycomb space tube (ST) where SVXII is held by 15 CF-glass legs e glued fc

We will refer to (C) central region where |rj| < 1 and to (F/B) forward (or backward) region 1 < |*j| < 2. c The expectation is that LOO wafers can last < 7.4 f t r 1 if kept at -5°C. «*Basic SVX active units: 1 wafer in the case of LOO, 2 wafers and 2 hybrids glued together and to a CF support for the SVXII and finally 3 wafers and 1 hybrid glued to a common CF support for ISL. "With low coefficient of thermal expansion.

122

to 3 adjustable mount points (differential screws) per barrel. LOO and the Be beam pipe are clamped to the ST via adjusters and dampers. Externally the SF is closed by a 0.1 cm thick CF-shell and connected to two CF-honeycomb extension cylinders (~9 kg each after detector cabling) that provide housing for the juction cards (and ISL port-cards). The extension cylinders rest on 6 inchworms (piezo-eleetric actuators) for the SVX active-alignment. The final weight of SVX, including cabling and cooling, is ~128 kg and the SF supports more then ten times its weight. In Figure 1 (left) we see the modules layout. Layers are concentric and within each one of them, modules are arranged so that some are closer to the beam line (inner) and others are further from it (outer)^. Nevertheless for SVXII, ISL the number of readout channels per ladder within a layer is fixed while, for LOO, space constraints led to a variable number (and corresponding use of wide and narrow sensors) (see also Table 1). The resulting geometry has in each layer some overlap regions that not only compensate for the inactivearea of wafer's borders but provide some thin double-hit ^-sectors (within a layer) that can be used to restore good resolution of tracks intersecting wafers ^-edges.

Figure 2. (left) LOO hybrid installation (low-mass cables visible); (center) Module installation under Coordinate Measuring Machine (CMM) during SVXII barrel assembly; (right) Side view of ISL layers 6 central and 7 forward in the "rotation-cage" used for module installation.

Several kind of 300 /xm thick wafers from different manufacturers (ST Microelectronics, Micron and Hamamatsu) have been used. All SVX layers are double-sided except for LOO. In particular layers 3, 5, 6 and 7 have axialstrips (wafer np side) on the p side 9 with small-angle-stereo (SAS) strips (wafer z side; SAS form an angle \a\ = 1.2° with the CDFII longitudinal •^Only in ISL: inner modules have wafers rip side facing out of CDFII while outer ones have the rip side facing the beam-line (see below for a definition of the rip side). 9 Only for layer 6, p side is SAS while n side is rip.

123

SVX layer 0 LOO 1 2 3 4 5 SVXII 6C 6F/B 7F/B ISL

r (cm)

pitch (/mi)

in 1.35

out 1.62

rtp 25

z -

stereo angle (deg.) -

2.5 4.1 6.5 8.2 10.1

3.0 4.6 7.0 8.7 10.6

60 62 60 60 65

141 125.5 60 141 65

90 90 +1.2 90 -1.2

22.6 19.7 28.6

23.1 20.1 28.9

112 112 112

112 112 112

=Fl-2 ±1.2 =Fl.2

N module [tot] 36/36 [72] 36 36 36 36 36 [180] 28 48 72 [148]

%x0 avg. [max.] 0.6 [1.1]

7.0 [15.0]

1.0 [5.0]

Table 1. Summary of the SVX layout. In the last column we report an estimate for the average (and maximum) %X0 (for 90° incidence) transversed in each SVX subdetector 6 . Maximum values refer to (r, 2 GeV/c, \d\ < 50 fim and a level 1 prerequisite of at least 2 XFT tracks. Figure 2 (left) shows the distribution of the second largest impact parameter in the event Using this data (corresponding to ~ 15 n& _1 of integrated luminosity) a small signal of D° —> Kir was reconstructed (figure 2, right).6 "Additional, lower order, corrections can be applied: like correcting for a residual non linearity in d and <j>, and for the misalignment of silicon layers within a wedge. However their online implementation in the online is less straightforward because it requires reprogramming of some SVT boards and adjusting of the SVT map which describes the detector geometry. The effect of these additional correction has been studied, and found to reduce the sigma of the i.p. distribution from 48 to 45 /im.

133 CDF Trigger oa Impact

D° -» KM signalfromtrigger tracks

1.3

1.4

1.5

1.6

1.7

1.8

1.9

2

2.1

2.2

(KreXGeV/c2) Figure 2. Plots done with d a t a selected by the level 2 trigger requests described in the main text. Left: Impact parameter of the track with the second-highest impact parameter in each event. Right: Invariant mass distribution for track pairs assuming the two tracks to be n and K, and applying cuts optimized to select a D° —»• Kn signal.

4

Conclusions

The SVT is the new level 2 trigger processor dedicated to the reconstruction of tracks within the tracking chamber and the silicon vertex detector of the CDPII experiment. The device has been thoroughly tested during the early phase of Runll data taking (April-October 2001). The performance is already close to design. Jut.© S © i @ EtC €s§3

1. J. Spalding, Run~II upgrades and physics prospects, these proceedings. 2. S. D'Auria, Commissioning and Operation of the CDF Silicon Detector, these proceedings. 3. F. Palmonari, The CDF-II Silicon Tracking System, these proceedings. 4. C. Ciobanu et al, IEEE Trans. Nucl. Sci. 46, 933-939 (1999). 5. S.R. Amendolia et al, IEEE Trans. Nucl. Sci. 39, 795-797 (1992). 6. W. Ashmanskas et al, Performance of the CDF Online Silicon Vertex Tracker to be published in IEEE Trans. Nucl. Sci. (2001). 7. S. Donati et al, Nuovo Cimento 112A n . l l , 1239-1243 (1999).

COMMISSIONING A N D OPERATION OF T H E CDF SILICON DETECTOR S. D'AURIA ON BEHALF OF THE CDF COLLABORATION Dept. of Physics and Astronomy, University of Glasgow, G12-8QQ Glasgow, U.K. e-mail [email protected] The CDF-II silicon detector has been partially commissioned and used for taking preliminary physics data. This paper is a report on commissioning and initial operations of the 5.8m2 silicon detector. This experience can be useful to the large silicon systems that are presently under construction.

Introduction The collider detector at Fermilab (CDF) is designed to study protonantiproton collisions at a centre of mass energy of 1.98 TeV. The experimental apparatus had a major upgrade recently, in order to take advantage of the upgraded luminosity of the Tevatron accelerator. A general description of the apparatus can be found elsewhere in there proceedings 1,z and in the Technical Design Report 3 . The Silicon Detector is a high-precision microstrip tracker that covers the pseudorapidity region |^71 < 2, with an outer radius of 28 cm and a coverage of \z\ < 50cm along the beam axis, around the nominal interaction point. The detector consists of 3 subsystems: "LOO" is the innermost, single-sided strip layer at 1.35 cm from the beam line; "SVX" is a 5-layer, double-sided silicon microstrip tracker, made of 3 identical barrels placed along z, each with a 12-fold geometry along the polar angle (f>; "ISL" is the outer part 4 , consisting of one-layer, double-sided central barrel and two double-layer barrels, also with double-sided silicon microstrips, in the forward and backward regions. The three subsystems use the same front-end chip 5 ' 6 , the same control cards 7 and the same readout system 8 . They differ in the sensor geometry, the read-out hybrids and power supplies. The total area of the double-sided silicon is 5.8m2, and features 722500 strips read out by 5644 front-end chips. The chip is the heart of the Silicon Detector. It consists of a charge amplifier, a double correlated sample and hold circuit, an analog pipeline, a comparator and ADC and a threshold logic for sparsification of 128 strips. The analog pipeline is 42 cells deep and works at 7 MHz. The chip allows for Dynamical Common mode noise rejection (DCMNR) and dead-tilme-less

134

135

operation. Each chip is programmable with a 197 bit word to set bandwidth, signal polarity, DCMNR, threshold, calibration mask. The interconnection system includes 1.75 million wire bonds to single strips, 310 thousand wire bonds to chips, 10 thousand wire bonds to control lines and 816 connections to receiver/transmitter circuits that control more than one module (5 modules for SVX, 10 for ISL). Given the size of the detector and its flexibility in terms of parameters to set, the commissioning was a formidable task. 1

Commissioning

The Silicon Detector was extensively tested at the fabrication facility throughout the construction process. Each part had passed tests with very stringent parameters e.g. less than 2% of disconnected channels, no readout errors 9 . Functional tests were made at each step. All ladders had passed the tests before insertion in the barrel, but after barrel assembly 11 modules out of 360 have developed anomalously high noise on single channels or clusters of channels. We believe this problem is due to buildup of surface charge on the interface with the oxide layer. In fact it affects only a small fraction of detectors and only in layer-2 and layer-4. These are small-angle stereo detectors, their fabrication structure is different from the one of the remaining layers and they come from a different manufacturer. The detector was repeatedly tested during assembly and before shipment to the experimental hall, which is located a few kilometers away from the fabrication facility. The system grounding was always reasonably good and, provided that all the ground straps of the ladders were connected to the bulkhead, the noise performance was, for the majority of the devices, the same as measured on single devices before assembly. We could not test the Silicon Detector in the assembly hall because the electronics and power supply crates are mounted on the walls of the collision hall, while the cables had been installed on the main CDF detector. So the complete chain of readout, power supply and controls had to be tested all at the same time when the detector was rolled in. We initially cabled only a part of the detector consisting of 50 ladders. We finished the cabling during a one month shutdown of the accelerator. We tested parts of detectors as long as they wer cabled. The cables, the power supplies and the data acquisition components had been previously tested separately with one standard test stand and had passed the specification requirements. We repeated a detailed test of each component when it was in place in the collision hall. The power supplies and their cables were first tested

136

with a resistive load and voltages, currents, protection circuits and interlocks were checked. Then functional tests of the DAQ-related part were performed making use of a portable test device, that we called "wedge in a box". It consists of 5 SVX hybrids and one portcard, with solid-state cooling. This device was known to read-out correctly within a defined range of conditions so that functional test and debugging was performed on the DAQ and on its cables. This method was also essential to identify 3 low voltage cable bundles that had passed the passive load test but had a dispersion on the ground connection, due to mechanical damage. Once the "outer" part of the system was fully debugged, the real detector segment was connected to it and completely tested. The cabling and testing operation was long in time and required a large effort for a variety of reasons: the space available in the bore and the clearance for plugging the cables were extremely limited; the failure rate of component that were previously was higher than expected. It required a large, organized effort by 30 people. The main operating difficulty was due to the tuning of the interlock system that protects the system and was being commissioned at the same time. It had to allow the system to be powered under non normal conditions, especially high humidity, and keep the detector safe. A false alarm on the temperature was, initially, difficult to recognize and could block the cabling crew for several hours, inhibiting any power to the detector. Cables are assembled in bundles consisting of one voltage cable, one sense, one command, one bias voltage cable and 5 optical cables. A small number of cables had to be replaced: 4 voltage cables, 3 voltage sense cables, 6 data (optical) cables, 3 command cables out of 114 bundles. Also 3 FIBs (Fiber Interface Board) had to be replaced. The largest difficulty was due to the optical link between the front-end (portcard) and the Read-out (Optical Fiber Transition Module). The high failure rate was due to the light level mismatch between transmitter and receiver. A 9-channel optical cable is driven by a custom-made monolithic DOIM GaAs laser 10 . Five transmitters, (45 channels) share the same power voltage. They have been selected to have about the same characteristics, but the same was not done for the receivers. Some difference in light level produced a considerable error rate. In addition some receiver modules had flaky pin connections to the VME board. Re-insertion after contact cleaning was necessary for a large number of receivers. Another source of difficulty was due to the power supplies n : 35 out of 102 developed problems and had to be repaired. We had not received the PS modules in time for detector commissioning and were only able to operate a partial system. We have commissioned and operated 70% of SVX and 35% of

137

ISL. The LOO is only partially commissioned due to the late arrival of power supplies and also to wait for stable operation of the Tevatron beam. The ISL had a cooling blockage problem 2 , but the totality of the detector has been functionally tested, but only for a very short period to avoid overheating. The overall result of this test was that 3 SVX wedges could not be readout. For one of them the problem has been identified in a short between two signals inside the detector. This was not present before shipment. As a lesson learned, at level of system design, we should have allowed the use of standard protections for vital wire bond connections, at the cost of a more difficult procedure for test and re-work. Also minimizing the transport of the sealed detector, if at all possible, would be desirable. Two other wedges are being investigated during access in October 2001. 9 wedges could not be powered due to lack of tested power supplies, 48 ladders had readout problems related to optical power mismatch and could not be operated. 2

Integration

After test in the collision hall, the ladders were re-tested one by one with a stand-alone version of the DAQ program, checking for readout errors under "standard" conditions. Ladders that passed this test were integrated in the CDF DAQ. We operated 70% of the SVX and 35% of ISL. We had some difficulties also in this phase. Communication errors with the Power Supplies gave rise to spontaneous turning on and off of apparently random channels. This problem was solved by improving the timing of this communication, by decreasing the number of power supplies served by the same serial line and with software checks. In particular the PS Users Interface program was charged to add a number of tests that made it considerably slower than foreseen. As the light output of the DOIMS depends on their operating temperature, we had new cases of mismatch due to the increased light power when operating at - 6 ° C . We had to develop tools to synchronize the power supply to the daq system and to constantly monitor the operation in order to respond efficiently to any error message. The number of ladders integrated vs. time is shown in fig. 1 (a). The steady linear increase was due to the testing procedure, that had to negotiate time with "physics" data taking of the rest of CDF, so that we could operate the detector only in a fraction of time. In addition, we required the Silicon to be in off status during beam injection and unstable beam conditions. The availability of DAQ time was the main factor limiting a rapid increase of the number of integrated ladders. The setback around

138

Figure 1. (a) Number of Silicon modules integrated with CDF vs. time, (b) Timing pulse height vs. relative phase of the chip clock with respect to the beam crossing.

day 260 was due to a VME power supply failure and momentary inability to operate a part of the system. 3

Operations

We have collected to date 4.5 ph~l of Physics quality data on tape; this excludes detector studies and special runs to check the trigger rate. In order to operate the Silicon in the most stable way we decided to make full use of the chip capability and operate it in DCMNR-on mode. A fixed threshold of 5 ADC counts, about 16% of the most probable m.i.p. signal, gave a reasonable readout time, and occupancy completely acceptable at the present low trigger rate. We have not optimized the chip parameters yet, although we are already using the best compromise between having a uniform standard set of parameters and good performances. The first variable to set in the Silicon system was the timing, i.e. the relative phase between the bunch crossing and the issue of LI trigger to the chip. Failing to syncronize correctly would result in a loss of charge and "spillage" of charge in the neighbouring beam crossing packets. Before plug-in we have measured the delay of all command cables. They were all the same, within 2 ns. Using the the data from the first beam collisions we did a coarse and fine time scan, as shown in fig. 1 (b), and verified that all the detectors show a maximum at the same delay, as expected. The noise performance with and without beam are as expected and we are not experiencing any measurable pick-up from the beam or from the outer part of the detector. Some ladders have a 50% increased noise in those channels located above the support rails. This sensitivity to the infrastructure is

139

probably due to loosened ground connections and the problem is being-addressed during access. Otherwise the noise is the same as measured at the fabrication facility. Measuring the pedestal and noise is essential to operate correctly the detector. Two calibration methods have been implemented. Firstly, in "Datamode" , data are collected with free running trigger in read-all mode. Then noise and pedestals are calculated off-line and the parameters written to the calibration database. This method is intrinsically slow, because the calibration constants are available a few hours after the run is finished. Especially in the initial commissioning phase, when detector parameters and configuration were changing continuously, we experienced difficulties in monitoring the data quality due to the time delay between data taking and calibration data ready. The "X-mode" calibration will allow for a fast turnaround. All

la N

M m

\

M»

Layer 0

IS

Signal Peak 22.8

m

Strip Noise Width 1.7

I JL I «^fc. »

m m

>

" »

:

' " ' < * ' • •

m

m

CA»C

so

ioo

iso

a»

300

Figure 2. (a) Signal to noise for clusters on tracks, as measured with one ladder that was read in read-all mode. (b)Correlation between the cluster charge on z side and cluster charge measured on the <j> side, for all clusters. No tracking is applied. The gain of the amplifiers are well equalized.

the calculations of pedestal and noise are performed by the VME cpu in the collision hall. The final result is then loaded in the database in a matter of minutes. The FIB module can also subtract on-line any residual pedestal, so that data will be ready for clustering and for taking part to the second level trigger (SVT) 12.

140

10

15 20 2S m 3S 40 43 SO m&t Prsba&te duster Chsrge $0G counts*

Figure 3. Distribution of the peak of the Landau distributions. There is an entry in this plot for each ladder that was read-out. The dashed histogram refers to the dimuon mass peak: in fig. 4 the distribution with Silicon in is compared with the one obtained using only the information from the central tracking chamber and from the muon central chambers: the width is only 20% better, but the S/N improved by a factor of 2.5. The Silicon information has also been used in the second level tracking trigger 12 .

•)/>•> M V ,

-

r

|

-•

|

i

••••

1

'

l]

S

i I

1

• i l i ' i I '— •i Si J-V 434*24 tfWflfs. 3^2]±0.01 M*jVVC"

-

'I

-

£ 2000 _ -

|

"

1

f

1

1

V

.

.„„«-"- 1

I

150

83 g

100

*s 1

3.2

,

1

3.*

,

1

.

3.6

1 3.S

^•v****

.

4.0

2.70

2.90

3.10 3.30 HH Mass GeV/'c*

Figure 4. The J/ip peak as measured requiring two muons in the central region. On the right it is shown the spectrum for the subset of events that have at least one muon as a Silicon track.

Conclusions The CDF Silicon Tracker is on is way to be fully commissioned. 70% and 35% of the SVX and ISL subsystems respectively have been taking data with beam collisions and have been used for extracting the first Physics signals. After an access to the detector that is occurring during this conference we

142

plan to be able to run 97% of SVX wedges and recover as much ISL cooling as possible. The detector will be fully operational for physics-quality data taking by January 2002. Acknowledgments We thank the Fermilab staff and the technical staff of the participating institutions. This work was supported in part by Particle Physics and Astronomy Research Council, the U.S Department of Energy, Istituto Nazionale di Fisica Nucleare, The Ministry of Science Culture and Education of Japan and Academia Sinica, Republic of China. References 1. J. Spalding, Run-H upgrades and physics prospects, these proceedings. 2. F . Palmonari [CDF Collaboration], The CDF-II Silicon Tracking System, these proceedings. 3. CDF Collaboration, The CDF-II Detector Technical Desiggn Report, FERMILABPub-96/390-E. 4. A. Affolder et al. [CDF Collaboration], Intermediate Silicon Layers Detector For The Cdf Experiment, Nucl. Instrum. Methods A 4 5 3 , 84 (2000). 5. M. Garcia-Sciveres et al., The SVX3d Integrated Circuit For Dead-Timeless Silicon Strip Readout, Nucl. Instrum. Methods A 4 3 5 , 58 (1999). 6. T. Zimmerman et al., SVX3: A deadtimeless readout chip for silicon strip detectors, Nucl. Instrum. Methods A 4 0 9 , 369 (1998). 7. J. Andersen et al. The portcard for the Silicon Vertex Detector Upgrade of the Collider Detector at Fermilab IEEE Trans. Nucl. Sci. 4 8 , 504 (2001). 8. M. Bishai [CDF Collaboration], The CDF Silicon data acquisition system for Run-H, these proceedings. 9. G. Bolla [CDF Collaboration], Testing And Quality Insurance During The Construction Of The Svxii Silicon Detector Nucl. Instrum. Methods A 4 7 3 , 53 (2001). 10. M. Chou et al.Dense Optical Interface Module (DOIM)Fermilab Internal document, Mar. 1996 11. Custom made power supplies for the SY527 system are made by CAEN. 12. I. Fiori [CDF Collaboration], The CDF on-line silicon vertex trigger, these proceedings.

T H E ASSEMBLY OF T H E A M S SILICON T R A C K E R , VERSION 1 A N D 2 C. CECCHI University

of Perugia

and INFN, Via A. Pascoli, 06100 Perugia, E-mail: [email protected]

ITALY

The AMS (Alpha Magnetic Spectrometer) experiment is a detector designed to search for antimatter and dark matter. A first version, AMS1, has flown on June 1998, on board of the Shuttle Discovery, during the STS91 mission. The complete detector, AMS2, will be installed on the International Space Station in 2004 and it is foreseen to operate for a period of three years.

1

Introduction

The AMS experiment is a space born detector which will search for antimatter and dark matter by measuring with the highest accuracy the Cosmic Rays composition, thanks to its large acceptance (~ 0.5 m2sr) and long observation time (three years). In this paper I will give a short overview of the AMS experiment 3 ; first I will describe the construction of the tracker of the AMS01 detector and I will then present the status of the construction of the new silicon tracker for AMS02. The performances achieved with the old detector, and the expected results with the new one, will be also discussed.

2 2.1

AMS01 The AMS-01 detector

The detector 4 is composed by a permanent magnet equipped with a tracker, which consists of 6 planes (two outside the magnet and four inside) of 300 /jm thick double sided silicon microstrip detectors, a time of flight system, based on scintillation counters, to measure the velocity of the particles, and a threshold cerenkov counter to discriminate between low energy hadrons, electrons and positrons. The magnet is made of blocks of Nd-Fe-B and gives a dipolar field of 0.14 T. The total acceptance of the detector is 0.5 m2sr, for an analysing power of 0.14 Tm2. The spatial resolution of the tracker is 10/xm in the bending plane and about 30 ^m in the non bending plane 5 .

143

144

'

..

.MffFLDINGSUrPOIU (AliOX)

HYBjiUDEOXdOP) 10HG X M T O N

CF1SHYBEBD L«J>DlRHSIHIOH.-::.Mi!.w l

>..«" HYBRID BOX (BOTTOM)

Figure 1. View of the components of a ladder (left). View of one ladder, silicon side (right).

2.2

The assembly of the AMS01 tracker

The basic component of the AMS tracker is the so called "ladder" (a detailed description of the ladder assembly procedure can be found in *). The ladder is made of several silicon sensors (from 7 to 15) aligned and glued together. A foam reinforcement is glued on both sides of the silicon. On one side of the ladder is glued a upilex fanout to reroute the strips signal to the readout electronics, which is placed at the end of the ladder. A view of the components used to assemble a ladder is shown in Figure 1 (left). The ladders are then mounted on Carbon Aluminium honeycomb. A picture of the ladder in the bending side is shown in Figure 1 (right). To obtain a high quality detector, two aspects have been particularly considered: stringent requirements on the quality of single components and mechanical precision of the assembly procedure. For the first one, acceptance criteria have been applied on silicon sensors and on the ladder itself in order to fullfiU the final specifications. The most important parameters used to accept or to reject sensors and ladders are shown in Table 1. In total 65 ladders have been produced, 6 of them have been rejected because classified as bad. 21 ladders out of 65 were classified as marginal, meaning that they had one or more parameters close to the edge of the acceptance criteria. Ladders can be declared marginal or bad because of different

145 Table 1. Acceptance criteria applied on silicon sensors and on assembled ladder for AMSOl and AMS02. Silicon acceptance criteria Total leakage current Hot strips s-side Hot strips n-side Ladder acceptance criteria Total leakage current Hot strips s-side Hot strips n-side

AMSOl

AMS02

6/JA

2nh

8 15 AMSOl 30/zA 3% 5%

6 4 AMS02 10M 2% 2%

reasons, assembly problems, damaged sensors, gluing problems (insufficient glue, overspread), bonding problems (damages to substrates, failed bonds), high leakage current or bad electronics. At the end of the construction of the AMSOl silicon tracker the number of bad channels, on all the six planes was of the order of 9%. The second important point of the construction of the tracker is the accurate positioning of the sensors in the ladder. The alignment relies on a very precise cut of the wafers. Therefore the first step before the assembly is the check of the sensor cut. The distance between the reference crosses of the wafers and the edge of the sensor has been measured for all the wafer and a precision of the order of 3-4 /xm has been found. The performance of the AMSOl silicon tracker, in terms of the momentum resolution, which is strictly related to the precision in the alignment, is shown in Figure 2.

3 3.1

AMS02 The AMS02 detector

The future AMS02 detector will consist of a superconducting magnet equipped with a tracker, which consists of 8 planes (two outside the magnet and six inside) of double sided silicon microstrip detectors, of 300 ^m thickness, for a total of 7m2 of silicon detector. A time of flight system is present, as in the prevoius version, to measure the velocity of the particles. The apparatus is completed by a transition radiation detector to separate electrons from protons up to 300 GeV, a ring imaging Cerenkov detector to study heavy nuclei, and the an electromagnetic calorimeter to measure electrons and photons up to 1 TeV. The superconducting magnet gives an analysing power of 0.9 Tm2, a factor of seven more than AMSOl.

146

I

• Test Beam Data I STS91 Data

•

[ D

Figure 2. Resolution of the tracker of the AMS01 experiment as function of the rigidity as measured in the test beam(dots) and during the flight(flight).

3.2

The construction of the AMS02 silicon tracker

The AMS02 silicon tracker is made of 192 ladders of different lengths. The components for the assembly of the tracker are at an advnced phase of production: Silicon sensors Qualification and preproduction phases have been terminated. The production yield is of about 70%, taking into account the specifications required on number of hot strips on p-side and on n-side and on the total leakage current. Upilex fanout The first preproduction is finished, the second one is on progress. A total yield of 50% on K5 (long cables) and of 85% on K6 (short cables) has been produced, the yield been limited mainly due to bonding failures. Front End electronics 6 Production of capacitors and front-end hybrids is in progress, with a yield of about 80 % on both of them. The final goal is to produce 192 ladders, plus 30% spares, totally equipped and functional. The AMS02 silicon tracker is being assembled in three assembly centers, Pe-

147 160 140 120 100 80 60 40 20 0

s

£

1

60 50

ovn.«

TOO

^

40 30

L 2

20 10 UVJ . . . . L.. 4 6

. . I ... .

0

8

L

r

250

i w u . ,M ., 2

4

...., .

6

noise distribution k

noise distribution S

L™

140

-

£

"S,

120

200

100 150

80 60

100

40 50 0

20

,,

.

i

20 40 w noise distribution S

.

0

1.,

.

1

.

20 40 law noise distribution k

Figure 3. Noise measured in one of the produced ladders. The bottom plots show the noise measured without subtracting the common noise.

rugia, Geneva and an italian industrial research center G&A Engineering 2 . Due to the relatively large scale of the construction, an industrial approach has been chosen. This has several implications; in particular a stricter documentation is necessary as well as a strong interaction between physicists and the company. A large effort has been put in the technology transfer to the industry. The industrial research center G&A Engineering, will assemble half of the ladders, the rest will be shared between the University of Perugia and the University of Geneva. A total of 27 ladders have already been assembled at the rate of 1.5 ladder/week. The results from the first 6 ladders are available. An example is shown in Figure 3, were the noise measured in one of these ladders in shown. The results of these tests suggest that the quality of the AMS02 ladders is excellent. The future plans for the AMS02 tracker construction are organised in order to complete the ladder assembly by the end of 2002; after that the assembled ladders will be integrated in the tracker, and finally the complete test of the detector will follow. The flight for the AMS02 detector on the International Space Station is foreseen for end 2004, beginning 2005.

148

4

Conclusions

The AMS01 silicon tracker comprised 57 ladders, for a total of 2m 2 of silicon microstrip detectors. A lot of experience has been gained in assembly procedure and a good quality has been obtained in assembly accuracy and in module quality. The AMS02 silicon tracker is now in construction, and an industrial approach has been choosen, due to the dimension of the detector. The assembly is in progress in three different assembly lines. A total of 27 ladders have been assembled and the end of the assembly is foreseen for December 2002. References 1. 2. 3. 4. 5.

Talk of M. Pauluzzi at Vertex 2000. G&A Engineering s.r.l., Localita Miole 100, 67063 Oricola (AQ)-Italy AMS Collaboration, J. Alcaraz et al., Phys.Lett.B461 (1999) 387-396 G.M. Viertel and M. Capell, Nucl. Inst. Meth. A419 (1998) 295-299 J.Alcaraz et al. Nuovo Cim. 112A 1325 (1999) W. J. Burger, Nucl. Inst. Meth. A435 (1999) 202 6. G. Ambrosi, Nucl. Inst. Meth. A435 (1999) 215

T H E AMS I N F R A R E D T R A C K E R A L I G N M E N T SYSTEM FROM STS91 TO ISS W. WALLRAFF AND V. VETTERLE /. Physikalisches Institut, RWTH-Aachen, Germany E-mail: [email protected] J. VANDENHIRTZ LemnaTec GmbH, Schumanstrafie 18 Wurselen D52146 E-mail: [email protected]

Germany

We report on AMS tracker alignment control in space using artificial laser produced straight tracks (flight data AMS-01, laboratory tests AMS-02) as well as precisely measured high momentum cosmics tracks.

1 1.1

AMS experiment AMS-01

The large acceptance Antimatter Spectrometer (AMS) experiment l 2 has been operated successfully on the NASA STS91 shuttle flight (02-June-98 12-June-98, AMS-01). It will be redeployed, including major upgrades, for a 1000 day data taking mission (AMS-02) on the International Space Station late in 2004; see talk by R. Battiston at this conference.

1.2

AMS Si-tracker Tracker Alignment System TAS

AMS particle tracking is based on 8/6 (AMS-02/01) planes of double-sided Si detectors providing a maximum detectable rigidity (MDR) of 3000(500) GV by measuring the sagitta of the tracks in a 0.9 T superconducting (0.12 T permanent NdFeB) magnet. The sagitta can be determined with an accuracy of 22(25) fim. In AMS the position stability of the tracking elements is controlled using nearly straight tracks. Fig. la shows the laser beams and their measured profiles (recorded in space and transmitted to ground on June 4th 1998) in the AMS-01 configuration. From an analysis of the residuals for > 4 GV tracks individual ladder displacements have been derived 3 4 (for principle see fig. lb, results fig. 4).

149

150

AMS Laser & Cosmlcs alignment

Figure 1. a) AMS 01 Si tracker and t h e Tracker Alignment System. T h e insert shows laser profiles observed while AMS was in orbit, b) ladder displacement measurement with cosmic tracks (curvature greatly exaggerated, 10 GV sagitta 0.5/3 m m for AMS-01/02).

1.3

TAS technical aspects

Artificial tracks are produced by 1082 nm Laser radiation. Si is highly transparent at 1082 nm, provided the natural reflectivity (nsi = 3.3) can be reduced and shadowing by the metallization of the readout strips can be kept small. AMS alignment sensors are antirelective coated and use 10 /im wide readout strips in the Laser impact areas. Thus single layer transparency can be as high as 50%. It has been shown (AMS-01) that a Laser ray can be recorded in 6 Si layers in sequence 4 . The AMS-02 tracker (8 planes Si, SC magnet) will be equipped with 2 sets of 10 laser rays each, that traverse the Si in 2 opposite directions (fig. 2b) and do overlap in the central planes. These rays are detected by generating electron hole pairs in the fully depleted Si particle detectors 4 6 . Signals from the alignment rays are recorded exactly like the charged particle tracks. 1082 nm Laser radiation is generated with high efficiency in DBR-Laser diodes coupled to monomode optical fibres that deliver - via miniature projection optics - low divergence circular rays into the tracker (fig. 2a). At the photon intensities readily available from Laser diodes (> 108 / pulse) signals exceeding that of 1000 mips can be produced in the Si layer (thickness 300 /im) close to the projection optics. At adequate Laser intensities this approach allows high precision (< 2 fan) tracker stability tests in very short time (< I s ) . The fully operational system (20 beams) weighs less than 5 kg.

151

AMS-02 S i - t r a c k e r &. l a s e r alignment rays 5 pairs chmck {x,y} in tke

cmntra.1 area {300 as 3.00 mm2}

mZTnm

of a l l 8 detector plaaas

Figure 2. AMS 02 Tracker Alignment System, a) laser radiation generation, b) basic configuration of 2 x 5 laser ray pairs.

2 2.1

TAS a n d A M S t r a c k e r stability STS91 flight

Overall the AMS-01 tracker 7 has been extraordinarily stable. Over the whole flight - including lift-off and landing - all tracking elements were found at their expected positions within ± 15 fun. Laser measurements (once per 3 orbits (on manual command)) were confirmed by observation with stiff (almost straight) tracks comparing extrapolated tracker hits with actually measured ones (fig. 4). Correcting for the time evolution of the small but finite displacements results in an approximately 20corrections are important only for the high rigidity tail of the cosmic ray spectra observable by AMS (p > 80 GV). The excursions observed are probably due thermal effects because they correlate with changes in flight attitude hence heating by the sun.

152

Figure 3. a) AMS-02 Si sensor transparency, standard and with antirefiective coating, b) high quality coatings eliminate front back interferences and distortions of the laser beam while passing through a sensor.

2.2

prospects

Based on AMS-Q1 experience a tracker stability verification along 10 lines and with better than 4 jura accuracy can be expected from short (< 10 s) runs 4 times per orbit. This measurement over an area of 300 x 100 mm 2 in the center of the acceptance is complemented by minimizing for high rigidity tracks over the full acceptance the pulls in the redundant trackfit with 8 points through the rather smooth and very stable AMS-02 B-field. The method of position control of Si trackers with artificial laser generated straight tracks has not only applications for space experiments. A similar system has recently been studied for implementation in the large area Si tracker of the CMS experiment 5 to be installed at LHC.

153

Figure 4. AMS 01 tracker stability during the STS 91 space flight; a) time line of y displacements (JLB), from stiff cosmic tracks; squares indicate Laser data. Frequency distributions of observed displacements in the AMS-01 (Laser) alignment ladders before b) and after correction c) observed for high momentum cosmic rays during the STS91 spaceflight; details are given in the references 3 and 4.

Acknowledgments NASA, DoE and DLR have generously supported this work. We like to thank the AMS collaboration and the Si tracker team for their cooperation. The meeting at Villa Olmo has proven again to be highly useful, many thanks to the organizers. References 1. 2. 3. 4.

U. Becker, ICRC XXVI, ice 1574, Salt Lake City (1999). W. Wallraff, JHEP-PREP-hep2001/211, Budapest (2001). W. Wallraff et al., ICEC XXVII OG 110, 2197, Hamburg (2001). J. Vandenhirtz Ein Infrarot Laser Positions Kontroll System fur das AMS Experiment, PhD thesis RWTH-Aachen (July 2001). 5. B. Wittmer The Laser Alignment System for the CMS Silicon Microstrip Tracker, PhD thesis RWTH-Aachen (November 2001). 6. Weihua Gu Characterization of the CMS Pixel Detectors, PhD thesis RWTH-Aachen (October 2001). 7. J. Alcaraz, et al.; A Silicon microstrip tracker in space; Experience with the AMS Silicon tracker on STS-91, Nuovo Cimento 112A, 1325 (1999).

P E R F O R M A N C E OF T H E BABAR TRACKER

SILICON V E R T E X

V. RE INFN-Pavia and Universita di Bergamo C. BOREAN, C. BOZZI, V. CARASSITI, A. COTTA RAMUSINO, L. PIEMONTESE INFN-Ferrara and Universita di Ferrara A.B. BREON, D. BROWN, A.R. CLARK, F. GOOZEN, C. HERNIKL, L.T. KERTH, A. GRITSAN, G. LYNCH, A. PERAZZO, N.A. ROE, G. ZIZKA Lawrence Berkeley National Laboratory D. ROBERTS, J. SCHIECK University of Maryland E. BRENNA, M. CITTERIO, F. LANNI, F. PALOMBO INFN-Milano and Universita di Milano L. RATTI, P.F. MANFREDI, INFN-Pavia and Universita di Pavia C. ANGELINI, G. BATIGNANI, S. BETTARINI, M. BONDIOLI, F. BOSI, F. BUCCI, G. CALDERINI, M. CARPINELLI, M. CECCANTI, F. FORTI, D. GAGLIARDI, M.A. GIORGI, A. LUSIANI, P. MAMMINI, M. MORGANTI, F. MORSANI, N. NERI, E. PAOLONI, A. PROFETI, M. RAMA, G. RIZZO, F. SANDRELLI, G. SIMI, G. TRIGGIANI, J. WALSH INFN-Pisa, Universita di Pisa, Scuola Normale Superiore di Pisa P. BURCHAT, C. CHENG, D. KIRKBY, T.I. MEYER, C. ROAT Stanford University M. BONA, F. BIANCHI, D. GAMBA, P. TRAPANI INFN-Torino and Universita di Torino L. BOSISIO, G. DELLA RICCA, S. DITTONGO, L. LANCERI, A. POMPILI, P. POROPAT, I. RASHEVSKAIA, G. VUAGNIN INFN-Trieste and Universita di Trieste S. BURKE, D. CALLAHAN, C. CAMPAGNARI, B. DAHMES, D. HALE, P. HART, N. KUZNETSOVA, S. KYRE, S. LEVY, O. LONG, J. MAY,

154

155 M, MAZUR, J. RICHMAN, W. VERKERKE, M. WITHERELL University of California, Santa Barbara J. BERINGER, A.M. EISNER, A. FREY, A.A. GRILLO, M. GROTHE, R.R JOHNSON, W. KROEGER, W.S. LOCKMAN, T. PULLIAM, W. ROWE, R.E. SCHMITZ, A. SEIDEN, E.N. SPENCER, M. TURRI, W. WALKOWIAK, M. WILDER, M. WILSON University of California, Santa Cruz E. CHARLES, P. ELMER, J. NIELSEN, W. OREJUDOS, I. SCOTT, H. ZOBERNIG University of Wisconsin, Madison The BABAR Silicon Vertex Tracker (SVT) consists of five layers of double sided, AC coupled silicon strip detectors. The detectors are readout with a custom IC, capable of simultaneous acquisition, digitization and reduction of data. The SVT is an essential part BABAR, and is able to reconstruct B meson decay vertices with a precision sufficient to measure time-dependent CP violating asymmetries at the PEP-II asymmetric e+e _ collider. The BABAR SVT has been taking colliding beam data since May 1999. This report will give an overview of the SVT, with emphasis on its running performance.

1

Introduction

The SVT design requirements and features are described in detail elsewhere 2 ' 3 . The SVT has been designed to provide precise reconstruction of charged particle trajectories and decay vertices near the interaction region. In conjunction with the BABAR drift chamber, the SVT is responsible for particle tracking. The design has been driven primarily by physics requirements, with constraints imposed by the PEP-II interaction region and the BABAR experiment. The PEP-II e+e~ asymmetric storage ring produces B-mesons couples at the T(45) peak with a boost j3j — 0.55 along the beam direction. The resulting average separation of B decay vertices along the beam direction is Az w 250/jm. To avoid significant impact on the CP asymmetry measurement, the mean spatial resolution on each B decay vertex along the z-axis must be better than 80 /mi 1 . To adequately recostruct B, r and charm decays, a resolution of order 100 /xm in the plane perpendicular to the beam line is needed. It is desirable that the SVT provide a tracking efficiency of 70% or more for tracks with a transverse momentum in the range 50 — 120MeV/c. This feature is fundamental for the identification of slow pions from D*-meson decays. The SVT is required to be able to withstand 2 Mrad of ionizing radiation. Forthermore, the accelerator environment demands a radiation monitor-

156

"- Beam Pipe

Figure 1. Schematic view of SVT: longitudinal section. The roman numerals label the six different types of sensors.

ing system capable of aborting the beams when detecting excessive radiation. Finally, the SVT must readout physics events at a LI trigger rate of 2000 Hz. Requirements and constraints have led to the choice of a barrel-shaped SVT made of five layers of double-sided silicon strip sensors. The modules of the inner three layers are straight, while the modules of layers 4 and 5 are arc/j-shaped (Fig. 1), to minimize the amount of silicon required to cover the solid angle, while increasing the crossing angle for particles near the edges of acceptance. To fulfill the physics requirements, the spatial resolution for perpendicular tracks must be 10-15 fim in the three inner layers and about 40 ^m in the two outer layers 2 . The inner three layers perform the impact parameter measurements, while the outer layers are necessary for pattern recognition and low pr tracking. 2

Performance

Due to a series of minor mishaps incurred during the installation of the SVT, nine out of 208 readout sections (each corresponding to one of two sides of a half-module) were damaged and are currently not functioning. There has been no module failure due to radiation damage. The SVT hit efficiency is measured by finding out if there are SVT hits corresponding to the traversed silicon sensors for reconstructed tracks. A global efficiency of about 97% is measured on the half-modules connected to functioning readout sections (Fig. 2). This includes inefficiencies from software reconstruction, dead channels, broken AC coupling capacitors, dead channels on front-end electronics and so on. The

157 1

1

& C

v.- ."""-r"" -

i

* • • ' •

i

1

' '

•

• "V

**-»** ;

+-*"-

--—.-„

- —_ ~

' • - . —

1

''"'"I

0.9

0.9

'o °-8 - 1

h

2

3

4

. 1

5

0.6

4

3

2

5

+

,

ti 0.7 I

i

.

1 . . . .

I !

20

30

+

,

,

i

i _ ^

,

.

: ,

i

• _•

20

40

forward half-module number

0.8

+

•

•

'

• '

30

•

, •

0.7 0.6

backward half-module number

Figure 2. SVT hit reconstruction efficiency, as measured on n+(i events. The horizontal axis corresponds to different modules, vertical lines separate different layers.

— 40

E 3 20 -50 0 50 angle (degrees)

-50

0

50 -50 angle (degrees)

0

50

-50

o

50 -50 angle (degrees)

0

50

Figure 3. Single hit resolution for (j> (left) and z (right) strips.

spatial resolution has been measured on tracks reconstructed by the SVT alone. The resolution (Fig. 3) depends on readout pitch, number of floating strips, noise, and is measured to be about 15/um and 30 — 40/xm for inner 1-3 and outer 4-5 layers, respectively. The reconstruction efficiency of slow pions from D* -¥ Dn decays has been estimated by comparing real and simulated data and found to be larger than 70% for pion momenta larger than 50 MeV/c (see Fig. 4). 3

Radiation Damage

A system of 12 PIN diodes is located near the first SVT layer to monitor continuously the radiation exposure of SVT and to protect the SVT from excessive radiation due to beam instabilities. The radiation dose strongly depends on the azimuthal angle: the diodes situated in the horizontal plane see about 10 times the radiation dose as the out-of-plane diodes. The highest measured dose at the time of writing (October 2001) is 920krad, to be compared with the radiation of budget at this time, 880 krad. Test detectors

158 .y 8000 — a )

. &

0.0

OJ. 02 03 0.4 Tiansveise M om entum G e V £ )

0i

a

01 02 03 04 Transverse M om entum PeVyfc)

F i g u r e 4. Slow 7r r e c o n s t r u c t i o n efficiency.

have been irradiated up to about 4Mrad and bulk type inversion has been observed at about 3 Mrad. The electrical properties after inversion have been measured and might allow regular charge collection;4 further investigations are ongoing to reach conclusive evidence. The front-end chips have also been irradiated up to about 4 Mrad: gain and noise degradation of about —3%/ Mrad and + 9 % / Mrad have been recorded, 5 respectively. According to current estimates, horizontal inner modules will accumulate a 2-3 Mrad radiation dose by the end of 2004, exhausting their planned lifetime radiation budget. 4

Conclusions

SVT has been operating efficiently since its installation in BABAR. The basic design goals have been fulfilled. Improved understanding on radiation damage on detectors and front-end electronics has been reached. References 1. The BABAR Collaboration, Letter of Intent for the Study of CP Violation and Heavy Flavor Physics at PEP-II, SLAC-443 (1994). 2. BABAR Technical Design Report, SLAC-R-457 (1995). 3. C. Bozzi et al., Nucl. Instrum. Methods A 447, 15 (2000). 4. I. Rachevskaia, Radiation damage to silicon by GeV electrons, talk given at the 5th International Conference on Large Scale Applications and Radiation Hardness of Semiconductor Detectors, July 4-6, 2001, Firenze. 5. A. Perazzo, private communication.

CHARGED PARTICLE TRACKING WITH THE HERA-B DETECTOR CARSTEN KRAUSS FOR THE HERA-B COLLABORATION Physikalisch.es Institut Universitat Heidelberg, Philosophenweg 12, 69112 Heidelberg The HERA-B experiment at DESY is a large acceptance fixed-target spectrometer using a silicon vertex detector, an inner GEM MSGC detector and an outer large volume honeycomb drift chamber for track reconstruction. The detectors are operated in a radiation environment comparable to LHC conditions. The tracking detectors had been finished at the beginning of year 2000 and have been successfully operated. They represent the worlds largest operated GEM MSGC system and the so far largest drift chamber system for high-rate application. We report on the detector operation, and summarize the performances achieved. We present the performance of the track finding algorithm and report on the reconstruction performance for the year 2000 data.

1

T h e H E R A - B Tracking S y s t e m

T h e H E R A - B detector was designed t o reconstruct decays of particles containing a b-quark with high accuracy. These particles are produced in proton (920GeV) nucleon interactions at t h e H E R A proton storage ring. T h e detector (see Fig. 1) is a forward wire target spectrometer with a silicon vertex detector, a large tracking system built of gaseous detectors, and particle identification detectors. T h e H E R A - B detector was operated until Aug. 2000, then repaired and upgraded during 2001 and will be ready for operation with

the HERA restart in 2002. T h e vertex detector system (VDS) is mounted in a vacuum vessel together with t h e target wires. It consists of 8 super-layers (each consisting of up t o 4 views) of silicon strip detectors with 50/xm readout pitch. T h e first 7 of these super-layers are mounted on R o m a n pots and can be moved to allow machine operation with increased aperture during injection. T h e particle rates in t h e VDS can reach u p t o 3 x l 0 7 s - 1 c m - 2 . In t h e center of mass system t h e coverage of t h e detector is larger t h a n 90% of t h e solid angle. T h e main tracking system of H E R A - B is divided into two p a r t s because the track density varies like 1/r 2 with distance r from the beam pipe. In the inner region (6-25cm distance from t h e center of t h e beam pipe) t h e tracks are measured with the inner tracker ( I T R ) . In this region the particle flux is u p t o 10 7 s - 1 c m ~ 2 . T h e inner tracker in t o t a l covers an area of 17m 2 . T h e outer region (20-600 cm radial distance from t h e beam-pipe) is covered by t h e outer tracker ( O T R ) . In this detector particle densities of u p t o

159

160

Figure 1. Top view of the HERA-B detector. The main tracking detectors stretch from the end of the vertex vessel to the electromagnetic calorimeter. The regions of the tracking system are labeled

10 s s' 1 c m - 2 are measured. The total active area covered by the OTR is 1000m2. The main tracking system starts behind the vertex detector. Several inner and outer tracker stations are installed in the magnet, where the chambers have to be able to work in a magnetic field of up to 0.85T. The tracking chambers are arranged in 0°, +5° and -5° stereo angles. 2

The Vertex Detector System

Charged particles produce more than 7 hits in the vertex detector which is sufficient for a stand-alone reconstruction. The tracks in the VDS are used to reconstruct primary and secondary vertex positions. This information can be used already on the second trigger level to cut on the distance of secondary vertices from the primary vertex to enrich the data sample with long-lived particles. The silicon detectors have a typical signal to noise ratio of 20-25 on the n-side and 15-18 on the p-side. The single hit resolution of the vertex detector is 3-4/Km. The vertex resolution has been measured to be around 40/jm. The very complicated alignment of the movable detector modules is stable and the positions of the system are known to a level of 2-7/mi in the direction perpendicular to the beam-pipe and 5G-250^m along the beam-pipe. This system is fully commissioned and reaches design levels.

161

Figure 2. a) Side view of the vertex detector vessel. The moving mechanism for the upper and lower pots can be seen. The protons enter the vessel from the right b) Side view of a single pot. The aluminum cover to separate the detector module from the primary vacuum is partially removed.

3

T h e I n n e r Tracker

The detector used in the inner tracker of HERA-B is a GEM MSGC. These chambers are a combination of MSGC 1 (micro strip gas chamber) and GEM 2 (gas electron multiplier), as shown in Fig. 3. Both devices produce gas amplification. At the GEM a gas gain of 20-50 and at the MSGC a gain of « 200 is reached. The division of gas amplification is necessary in the HERAB environment, because each device alone can not be stably operated in a hadronic beam with sufficient gas gain. The gain needed for the HELIX 128 readout chip 3 and a strip length of up to 25cm is at least w5000. The inner tracker system is built of 184 chambers with more than 140,000 MSGC strips. The gas used is a mixture of 70% Ar and 30% CO2. The chambers were designed for a maximal radiation exposure of 1 Mrad/y. The operation of gaseous micro pattern detectors in a hadronic environment is very difficult. It could be established that a careful conditioning of the GEM MSGCs in the beam is mandatory for a stable operation. Even after the training of the chambers, a slow switching-on procedure of the high voltage has to be strictly followed. In addition to the careful handling of the chambers, a fast monitoring system controls the high voltage and switches the system off in case of problems.

162

Figure 3. Schematic display of the GEM MSGC. The layout of an inner tracker chamber is shown on the right.

The most common operational problem caused by the conditions in HERA-B is a spark between the upper and the lower side of the GEM. These sparks can develop into discharges on the MSGC wafer or into shorts within the GEM itself. Both of the latter cases have to be avoided, because they permanently damage the chamber. The ITR has a GEM spark detection system integrated in the high voltage distribution system, which reduces the voltage applied at the GEM for « 1 minute after a GEM spark was detected. This protects the GEM from developing a permanent short from a spark. With all these measures it is possible to operate the GEM MSGC chambers in the high rate hadronic environment of HERA-B. On average the 150 inner tracker chambers installed in 2000 were operated for «1Q80 hours each. The inner tracker chambers had an average GEM spark rate of 1.2 sparks per chamber and 24h of operation. The GEM spark rate increases roughly exponentially with the applied voltage. The level of 1.2-8 sparks per operation day at voltages between 420 and 460V can be tolerated. The GEM voltages of all chambers were individually adjusted (420-460V) to reach a similar efficiencies. The individual adjustment was necessary to compensate the observed gain variations in the GEMs, which are most likely production induced. The efficiency level reached is shown in Fig. 4, efficiencies between 91% and 98% were measured, the design efficiency is 95%.

163 JL

•

' ~J

-

r

T—

"'"' ^^^^*^;;:-::;1"'

--''''

•

/'"'

„

~

±^

* — * MS91-+S < •-« MSM-+2 " »-•"-MS01-+3 ~ > - - * MS01-+4 "

'/

<tpm symbols i Udrift= -1A kV, Death. • .515 V

"~

H1W syratmls: UdrlS m -1J kV, Ucotfc » -SIS V

-

• — • MSM++I _ • — • MS81++2 .

/A'

7

[

,

1

.

1

,

1

,

1

,

1

GEMv«»ltsge[V]

Figure 4. Measured efficiency of some inner tracker chambers as a function of the applied GEM voltage.

4

The Outer Tracker

»)

K^E^^^33

Figure 5. a) Schematic view of the honeycomb chambers used in the outer tracker, b) Outer tracker efficiency for the chambers in the pattern recognition area with different T D C thresholds.

The outer tracker uses honeycomb drift chambers (Fig. 5a). These chambers are built of Pokalon Cfoil (a carbon loaded polycarbonate). This foil is coated with copper followed by a layer of gold to increase the surface conductivity. The gas mixture used in these chambers is 65% Ar, 30% CF4 and 5% CO2. The gas gain in the chambers is typically 30,000. Occupancies of up to

164 20% in the central region are measured. The module dimensions are typically 20cm width and up to 4.5m in length. The complete outer tracker is divided into 13 stations, which are instrumented with more than 1000 detector modules. During the R&D for the outer tracker severe aging problems had to be overcome. In the 2000 running period the system was fully installed and routinely operated, but production quality problems kept the efficiency lower than possible with these chambers. This can be seen in Fig. 5b, where some chambers with a high TDC threshold could only reach an efficiency of about 85%. Other - usually non trigger chambers - reached efficiencies of up to 95%. It will be possible to reach a higher level in 2002 because the problems have been identified and largely solved. 5

Track Reconstruction

The main tracking system of HERA-B is divided into three parts: The magnet area, the pattern recognition area and the trigger area (See Fig. 1). The track reconstruction starts in the pattern region. The reconstructed track segments are then extrapolated into the magnet and into the trigger area using a Kalman filter-based algorithm. The tracks are then matched to VDS tracks and a combined re-fit is performed to determine the momentum. During the commissioning phase in 2000 the detector performance was lower than expected and the reconstruction algorithm had to become more robust against missing layers. A new pattern recognition algorithm based on space-points was therefore developed in addition to the existing one based on triplet seeds. The space-points calculated from hits in several layers are matched and combined using quality criteria. This algorithm proved to be very robust and fast. In a Monte Carlo study with a very pessimistic assumption about the detector performance (85% efficiency for the ITR and 90% for the OTR and reduced resolutions for both detectors), the reconstruction efficiency for a /i-track from a J/ip decay is 97%. In Fig. 6 a typical reconstructed event in the pattern recognition region can be seen. The hits in the detectors are marked by crosses, the reconstructed tracks and track segments are marked by the interconnecting lines. 6

Summary

In HERA-B, the first tracking system for LHC-like conditions has been built. The high track density poses high requirements both on the detector technology and on the reconstruction. The vertex detector system is fully com-

165

missioned and works satisfactory. The GEM MSGC technology used in the inner tracker is difficult to operate in the given environment, but it is capable of Milling the requirements. The honeycomb chambers of the outer tracker are a suitable means to cover large areas in a high rate environment. The reconstruction is working with improved speed and efficiency. The commissioning of the HERA-B tracking system was largely completed in 2000. Due to improvements and repairs that have been completed in the meanwhile, the tracking system should reach design performance for the 2002 data taking.

Figure 6. A typical event in the inner part of the pattern recognition area of the main tracker.

References 1. A. Oed, Position Sensitive Detectors with Microstrip Anode for Electron Multiplication with Gases, Nucl Instrum. Methods A 263, 351-359 (1988) 2. F. Sauli, A new concept for electron amplification in gas detectors, Nucl Instrum. Methods A 386, 531 (1997) 3. W. Fallot-Burghardt et al., Helixl28-x User Manual, HD-ASIC-33-0697, http://wwwasic.kip.umheidelberg.de/~feuersta/projects/Helix/helix/helix.html, 1999

THE ZEUS MICRO VERTEX DETECTOR A. POLINI FOR THE ZEUS MICROVERTEX GROUP DESY, Notkestrasse 85, 22607 Hamburg, Germany, E-mail: [email protected] For the luminosity upgrade at the HERA ep collider, the ZEUS experiment has designed, constructed and recently installed a Silicon Microvertex Detector. The design of the detector and the performance of prototypes are discussed. The readout chain, the adopted online and control solutions as well as the integration within the existing trigger and acquisition systems of the experiment are presented. Tests with the full detector prior to installation and the first experience using cosmic rays in the final environment are reported.

1

Introduction

The ZEUS detector at DESY is designed to study high energy interaction produced at the HERA e±p collider. In the year 2000-2001, the 920 GeV proton - 27 GeV electron collider has undergone a substantial upgrade aiming at an increase by a factor 5 in the peak luminosity, corresponding to 200 p b _ 1 integrated luminosity per year. During the upgrade shutdown, ZEUS has been equipped with a silicon vertex detector which, besides a general improvement and extension of the track reconstruction, will enhance the identification of short lived particles. 2

Detector Layout

The Microvertex Detector (MVD) consists of a barrel section with three double layers of sensors surrounding the beampipe and four wheels in the forward, outgoing proton, direction. Longitudinal and transversal views of the detector with respect to the beam line are shown in Fig. 1. The sensors are single sided and made of high-resistivity (3 — 6 kti cm) 320 fj.ro. thick n-type silicon into which p+ strips, 12 /un wide and with a 20 fim pitch, are implanted. The signal is read out via AC coupling of 14 /mi strips placed at a pitch of 120 /jm. The rear side consists of a thick n+ diffusion. Test beam results have shown that, using capacitive charge sharing, a resolution up to 8 fjm can be obtained for tracks perpendicular to the sensor *. In the barrel region two consecutive sensors of square shape (60 x 60 mm), with orthogonal strips, are glued and electrically connected together via a copper trace etched on 50 pm thick Upilex foil. The connection of the

166

167 (a)

W

Forward MVD

0

I

D

Barrel MVD

[

=

(*>)

=

Figure 1. Transversal view with respect to the beam line of the forward (a) and the barrel sections (c) and longitudinal view (b) of the complete detector.

sensor assembly to the read-out hybrid is also done via a Upilex foil. This structure with a mirror one having perpendicular strip orientation forms a barrel module with 2048 read-out strips or 1024 channels. Five modules are mounted on a carbon fiber ladder that provides the required stiffness and support for the cooling pipes, cabling and slow control sensors. The forward section consists of four wheels, each made of two parallel layers of 14 silicon sensors of same type as the barrel section but with a trapezoidal shape and 480 read-out channels. Two sensors mounted behind each other form a forward segment and provide a two coordinate measurement via strips tilted by 180°/14 in opposite directions. A more detailed description of the detector layout and of the silicon sensors can be found in 2 . The read-out of the 207,360 channels is done via the HELIX 128-3.0 frontend chip 3 , specifically designed for the HERA environment. This 0.8 /xm CMOS chip is fully programmable and equipped with a 128 channel read-out system and a 136 step analog pipeline. The noise performance of the HELIX depends on the input capacitance (C) and is 400 + 40- C [pF] equivalent noise charge (ENC). Irradiation tests of the HELIX have been presented at this confererence 4 . During normal operating conditions the power dissipation of the HELIX is 2mW per channel while the power dissipation in the silicon sensors is negligible. Eight chips belonging to the same barrel module or the same forward segment are connected together in a programmable failsafe token ring and read out via a single digitization channel.. The analog serialized data are sent through passive copper links to dedicated 10 bit ADC modules 5 . The ADC system, after common noise and pedestal subtraction, provides strip clustering and two output data streams, the first based on cluster data for triggering purposes and the second for complete read-out of accepted events. The read-out is performed via VME Power PC boards running LynxOS and a dedicated software library 6 . The detector is operated at controlled humidity and temperature with dry air flow and water cooling infrastructure.

168

3

Final Assembly and Standalone test with Cosmic Rays

After assembly completion, the MVD was set up in a test environment and connected to the final read-out electronics and control infrastructure including the final set of power and signal cables. An extensive test program aiming at the complete understanding of the detector prior to the final installation was performed including a scope test of the front-end electronics and cabling, tests and development of the data acquisition and slow control system, study of signal to noise and detector performance under nominal conditions. Finally a cosmic trigger using scintillator layers surrounding the whole detector was set up and 2.5 million of cosmic events were collected over a period of 3 weeks of continuous running. During the test two barrel modules (out of 206) and a single front-end hybrid were found to be faulty. A preliminary analysis on the cosmic events has been performed requiring at least two hits in both projections in the modules of the outer layer. The input for these calculations was the designed geometrical position of the detectors. A resolution of 70 /im in r — (j> and of 80 fim in the r — z coordinate was obtained, dominated by the systematic uncertainty in the position of the sensors. After alignment corrections the design goal of a resolution of less than 20 fim looks feasible. Results from the cosmic run were also an average efficiency close to the geometrical acceptance of the silicon sensors (i.e. 0.93), a total number of bad channels lower than 2 % and a signal to noise ratio, including the full read-out chain, of at least 13. The noise and pedestal levels have shown to be stable at the level of 1-2 ADC counts although attention has to be payed to environmental parameters like temperature and humidity. Low humidity has turned out to be also important for low and stable currents when operating the sensors at depletion voltage. 4

Integration within the ZEUS experiment

In April 2001 the detector and the associated electronics infrastructure was moved to the final location in the ZEUS experiment. Quick tests on the whole detector have been performed as well as the integration in the ZEUS data acquisition and in the slow control environment. Before the HERA luminosity running, cosmic runs including the MVD and the other ZEUS detector components, were successfully achieved. 4-1

The MVD Data Acquisition and the new Global Tracking Trigger

The ZEUS data acquisition system is based on a three level trigger. Because of the HERA bunch crossing rate of 10.4 MHz, i.e. 96 ns between consecutive interactions, the experiments have required a pipelined read-out design. A

169 First Level Trigger, based on a reduced set of information from the detector components, is issued after 46 bunch crossings and reduces the trigger rate to aproximately 500 Hz. Detector data, stored in deadtime-free pipelines, is subsequently digitized, buffered and used for a Second Level Trigger (SLT) evaluation which is performed first by the single detector components, within 10 ms, and whose combined result is used to lower the rate to 100 Hz. For accepted events the complete detector information is read out and sent to the Third Level Trigger computer farm where event reconstruction and final online selection are performed. As a trigger component, the MVD has stimulated the realization of a new Global Tracking Trigger (GTT), based on the combined information of the Microvertex Detector together with the existing Central Tracking Detector (CTD) and the newly installed forward Straw Tube Tracker. The GTT will provide efficient trigger and reconstruction capability already at the SLT. For the hardware, the strategy has been to use, whenever possible, commercial off the shelf equipment easily scalable and mantainable. After investigation on the performance achievable in terms of data throughput and process latency, a solution based on a farm of standard PCs connected via a Fast and Gigabit ethernet network has been chosen. In the final system, data belonging to the same event and coming from the different tracking detectors is sent, according to a dynamic list of idle processors, to one computing node where the combined information is decoded and a complete tracking algorithm is run. After processing the GTT results are sent to the Global Second Level Trigger and the list of the idle processors is updated. All data transfers are done using standard TCP protocol. For development and performance tests a playback capability has been provided: upon a First Level Trigger the component front-end electronics (currently CTD and MVD) is read out and Monte Carlo or previously saved events stored in memory are injected into the GTT trigger chain at the component VME interfaces, and sent through the system exactly as for regular data. Preliminary measurements using a high transverse energy dijet photoproduction sample with high track multiplicity as well as previously recorded cosmic events, have shown no appreciable increase in the system deadtime or trigger latency, being consistent with the performance of the existing ZEUS DAQ and trigger systems during data taking at the end of 2000. Concerning the algorithm performance, the same simulations have shown that the track resolution is significantly better than the existing CTD SLT 8 , and either approaches or exceeds the current offline tracking resolution for tracks with only CTD information. The vertex resolution of the GTT system, even using only the CTD, is significantly better (10 mm) than the existing

170

CTD SLT (80 mm) due to the use of additional detector data not possible before because of the large processing time required. When including the MVD a nominal resolution of approximately 500 /j,m is reached. The efficiency of the GTT reconstruction to find the event vertex is similar to the present offline reconstruction, and approaches 100 % for events with 5 or more tracks. 4-2

Radiation Monitor

The MVD has been equipped with a composite radiation monitoring system to enable instantaneous and integrated dose measurements at several detector points and an automatic beam dump trigger. The designed lifetime of 5 years of operation in the HERA environment sets a limit on the integrated radiation dose that can be tolerated by the detector per year. A dose of about 250 Gy/yr can be seen as a safe limit for low signal to noise degradation partially recoverable by tuning of the front-end parameters. The system has proven to be an invaluable monitoring tool during the upgraded HERA machine commissioning. 5

Conclusions

The status of the new Microvertex Detector recently installed in the ZEUS detector has been reported. First tests in the complete environment and results based on cosmic data are encouraging. The Global Tracking Trigger concept looks promising and will significantly enhance the online tracking resolution. The luminosity run is expected to start early in 2002. Acknowledgements I would like to thank my colleagues from the ZEUS MVD group and in particular R. Carlin, U. Kotz, and C. Youngman for the useful discussions. References 1. 2. 3. 4. 5. 6. 7. 8.

M. Milite, DESY Thesis 2001-050 (2001). E. N. Koffeman, Nucl. Instrum. Methods A 473, 26 (2001). M. Feuerstack, Nucl. Instrum. Methods A 447, 89 (2000). J.J. Velthuis, these proceedings. T. Fusayasu, K. Tokushuku, Nucl. Instrum. Methods A 436, 281 (1999). A. Polini, ZEUS Note 99-071 (1999), unpublished. M. Sutton, C. Youngman ZEUS Note 99-074 (1999), unpublished. A. Quadt et al., Nucl. Instrum. Methods A 438, 472 (1999).

T H E R U N IIB U P G R A D E OF T H E CDF SILICON DETECTORS

S. CABRERA FOR THE CDF COLLABORATION Duke University, Physics Department, Durham, North Carolina 27708, E-mail:

[email protected]

A substantial portion of the Runlla silicon detectors of the CDF experiment will not perform adequately for the duration of Run l i b (15 f b - 1 ) because of radiation damage. The Silicon Vertex Detector (SVX-II) and Layer 00 will be fully replaced at the end of Run Ha. The Run lib silicon tracker has a baseline design that safely achieves the required radiation tolerance by using single sided sensors that are actively cooled. The new Run lib castellated layout contains more silicon surface area and has a more uniform radial distribution. It minimizes the number of hybrid and sensor varieties providing quick construction and assembly. The total mass in the tracking volume is reduced by eliminating unnecessary the passive material from the CDF volume.

1

Introduction

A successful Run II engineering run in 2000 established the 36x36 bunch pp operation at a center-of-mass energy of 1.96 TeV at the Fermilab Tevatron and produced (58 ± 17) nb~ of integrated luminosity for the comissioning of the CDF detector. Run II officially began in March 2001, and luminosity is currently being delivered to both the CDF and Dj? experiments which have started to carry out their ambitious physics programs 1 . Run II is divided into two distinct stages: Run Ha will collide proton on antiproton bunches with a bunch spacing of 396 ns switching to 132 ns bunch spacing at 2-1032cm~2 s - 1 of instantaneous luminosity, and Run lib will maintain the bunch spacing at 132 ns. The number of superimposed interactions per beam crossing a must be kept low enough to allow a comprehensive event reconstruction by the collider detectors. The latest prospects of the Run Ha luminosity levels are an integrated luminosity of 2 f b - 1 and a peak value in the instantaneous luminosity of 2-10 32 cm- 2 s- 1 . The transition from Run Ha to lib in the begining of 2005 will require a shutdown of approximately 6 months, primarily to replace the radiation dama T h i s magnitude obeys a Poisson distribution whose mean is a linear function of the instantaneous luminosity with a positive slope dependent on the number of bunches in the proton and p beam

171

172

aged components of the silicon detectors of the CDF and DJ? b experiments. The goal for the three year Run lib period (2005-2007) together with Run Ha is to accumulate 15 fb~ by increasing the instantaneous luminosity to 5-10 32 cm -2 s _ 1 . One of the major efforts is focussed on increasing the number of antiprotons per bunch in the collider by a factor of 2-3 over the Run Ha value of 3.0-1010 in 2 to 3 years without a major interruption to the Run Ha program. The upgrades in the accelerator system for p production, collection, handling and accumulation are well described elsewhere 3 . 1.1

Radiation damage in the Runlla silicon systems

The capabilities of the CDF experiment for the total integrated luminosity expected in Run II are limited by the radiation damage to the Run Ha silicon tracker c , which is estimated to survive to approximately 5 fb _ 1 . This radiation damage will affect both the silicon sensors and the readout electronics. The silicon sensors of the Run Ha silicon tracker will suffer a deterioration in performance primarily due to radiation damage to the bulk silicon, through displacements of silicon or impurity atoms from their lattice sites. One effect is an increase in the leakage current that degrades the ratio of signal to noise. A second effect is a change in the dopant concentration of the silicon which leads to an increase in the depletion voltage. The SVX-II sensors are double sided and must be operated fully depleted, otherwise the strips on the ohmic side would remain effectively shorted together. These sensors are inherently limited in the bias voltage that they can sustain: the electronics on both sides of the sensors are referenced to a common ground and the applied bias voltage must be held off by coupling capacitors that can withstand ~ 100 volts. The increased depletion voltage will be the dominant mechanism leading to the demise of the SVX-II layers with double sided sensors that combine axial and small angle stereo manufactured by Micron (layers 2 and 4). The same mechanism will be dominant in the case of the single sided Layer 00 sensors. The SVX-II layers with double sided axial and 90° stereo sensors manufactured by Hamamatsu (layers 0, 1 and 3) will die as a result of the combination of both bulk damage effects. The integrated luminosities that can not be exceeded in order to maintain reasonable detector performance were evaluated for the different components of the Run Ha silicon tracker 5 . These limits were under 15 fb _ 1 for Layer 00 and Layers 0,1,2, and 4 of the SVX-II. Other SVX-II components like b

See 2 for a complete description of the Run lib upgrade of the D 0 silicon tracker. See reference 4 for a complete description of the Runlla silicon tracker of the CDF experiment.

c

173

the Port-Cards 6 ' 4 ' 7 , that generate the control signals for the SVX3 readout chip, and the electronic components of the DOIMS (Dense Optical Interface Modules) connecting the Port-Cards with the front end electronics, are not expected to survive 5 fb _ 1 . All the layers of the ISL (Intermediate Silicon Layers) 4 will survive Run lib and will not need to be replaced. For mechanical reasons and the tight schedule for this upgrade programme a full replacement scenario of the SVX-II and Layer 00 systems was approved. 2 2.1

The baseline design for the R u n l i b replacement detector Comparison of the Run Ha and Run lib silicon systems

The baseline design of the replacement detector 8 ' 9 has been developed to accomplish several goals. In order to achieve the required radiation tolerance, single-sided highbias-voltage sensors will be used instead of the double sided sensors used in SVX-II detector. Therefore to retain or improve the tracking capabilities of the Run Ha silicon tracker at least twice as many sensors as in the current Run Ha detector will be needed. The current SVX3 chip will be replaced by a new rad-hard SVX4 chip. The new Run lib castellated layout (see figure 1) takes advantage of the entire volume between the beam pipe and the ISL space frame from 2 to 18.5 cm in radius. There is more silicon detector area given that the outer instrumented layers are located at larger radii in the Run lib design (see table 1). The mass distribution is better because the radial gap in the Run Ha detector between the ISL and the SVX-II which was occupied by passive material such as cables and Port-Cards (see section 2.3), has been eliminated. The wedge structure of the Run Ha detector, that maintained the same 12-fold <j> segmentation for all the layers, is abandoned in Run lib in favour of minimizing the number of hybrid and sensor varieties. This is required for the quick construction and assembly, and reduces the overall cost. Layers 2 through 6 (90% of the detector) have a uniform stave design and only 3 sensor types while SVX-II had 5 sensor types. There will be only 4 types of hybrids while SVX-II and Layer 00 have 12 hybrid types. The layers 0 and 1 together have 2 types of sensors, in comparison with the 2 types of sensors for Layer 00 alone in Run Ha. All the layers will use intermediate strips between the readout strips allowing smaller pitch and better hit resolution while keeping the channel count low. The total increase in the number of sensors from Ha to lib is ~250%

174 Power and Control Cables

Figure 1. Run l i b silicon tracker r-0 inner layout versus end view of SVX-II detector at the same scale. ISL detector is not included.

Table 1. The Run lib silicon tracker baseline design in comparison with the Run Ila SVX-II and Layer 00 detectors.

lib Layer L0 inner L0 outer LI L2 L3 inner L3 outer L4 inner L4 outer L5 inner L5 outer L6 inner L6 outer

lib axial R(cm) 1.95 2.35 3.35 4.55 6.45 7.70 9.50 10.6 12.5 13.6 15.5 16.6

lib stereo R(cm)

3.00 (90°) 4.90 (90°) 6.10 (90°) 7.35 (90°) 9.15 (2.5°) 10.25 (2.5°) 12.15 (2.5°) 13.25 (2.5°) 15.15 (90°) 16.25 (90°)

lib Ila 50cm) avoiding the problem of fitting hybrids on narrow inner sensors and reducing significantly the material and cooling requirements. The use of these cables causes a degradation in the signal to noise ratio by the noise pickup and a higher readout capacitance, and thus they are only used on layer 0. d e

See reference 6 ' 7 for more details about the Run Ila Port-Card. See 8 for more details about the Run lib Junction-Port-Card.

177

3

Conclusions

The CDF Run lib silicon tracker, which will replace the Run Ila detector because of radiation damage, has a baseline design which is comparable in performance and has many structural advantages. We have discussed this basic design and the components from which it will be built. The design includes 90° layers for precision tracking in 3 dimensions. We look forward to the physics this device will provide in Run lib. Acknowledgments I am indebted to the CDF Run lib Silicon design group, in particular I would like to thank P.Azzi-Bachetta, N.Bachetta, D.Benjamin, B. Flaugher, J.Goldstein, C.Hill, J.Incandela, M. Kruse, P. Maksimovic, T.Nelson, D. Stuart, K. Yamamoto and W-M. Yao for a very enjoyable collaboration. References 1. J. Spalding for the CDF coll., "The Run II upgrades and physics prospects'' , These proceedings. 2. A. Bean for the Dj? coll., "Design of un Upgraded Dj? Silicon Microstrip Tracker for Fermilab Run2", These proceedings. 3. M. Church, "Substantial upgrades to Tevatron luminosity", hepex/0105041. 4. F.Palmonari for the CDF coll., "The CDF II silicon tracking system", These proceedings. 5. N. Bacchetta et al., Run2b Silicon Working Group Report, CDF internal note 5425, September 2000. 6. M.Bishai for the CDF coll., "The silicon data acquisition system and front-end electronics for CDF Run II, These proceedings. 7. S.Dauria for the CDF coll., "Comissioning and Operation of the CDF SVX detector", These proceedings. 8. The CDF coll., "Run 2b Technical Design Report ", November 2001. 9. B. Flaugher and N. Bachetta (CDF coll.) internal talks: http://wwwcdf.fnal.gov/internal/run2b/Run2b-silicon.html 10. G. Bolla et al (1998-1999) "Silicon microstrip detectors on 6-inch technology", Nucl.Instrum.Meth. A435, 51-57. 11. T.K. Nelson, "The CDF Layer 00 Detector", Pub. Proc. The Meeting of the Division of Particles and Fields (DPF 2000) of the American Physical Society, Columbus, OH, Aug. 2000. FERMILAB-CONF-01/357-E

T H E B T E V PIXEL D E T E C T O R S Y S T E M L.MORONI * INFN Sez. di Milano, Via Celoria 16, 20133 Milano, Italy E-mail: luigi. moroniQmi. infn. it BTeV is a collider experiment approved to run at the Tevatron at Fermilab. The experiment will conduct precision studies of heavy flavor decays, with particular emphasis on CP violation, flavor oscillations and rare decays. One of its unique features is a state of the art pixel detector system, designed to provide accurate measurements of the decay vertices of heavy flavored hadrons that can be used in the first level trigger. This will insure the ability to perform precision study of a variety of final states and to search for rare phenomena with very high efficiency. The main design features of the pixel vertex detector are reviewed.

1

Introduction

BTeV is an experiment expected to be running in the new Tevatron CO interaction region at Fermilab. Its physics goals encompass precision measurements of all the CKM phases that can be extracted from CP violation observables in b decays, charm and beauty rare decays and flavor oscillations 1. This experiment exploits two important advantages of the "forward" region: the correlation in the direction of the b and b produced that improves the flavor tagging efficiency, and the boost that allows the implementation of an efficient trigger algorithm based upon the identifications of detached charm and beauty secondary vertices. The key element in this triggering approach is the pixel vertex detector. The pixel technology has been chosen instead of silicon strips because of the superior pattern recognition capabilities and stronger resilience to radiation damage. The geometry and readout scheme chosen are optimized for our physics goals through detailed Monte Carlo studies. A recent extensive beam run 2 that took place at Fermilab in 1999 validates the accuracy of our predictions. 2

BTeV pixel system overview

The pixel vertex detector is located at the center of BTeV, inside a 1.6 T dipole magnet. The detector system is located in a secondary vacuum, separated from the beam vacuum by a thin Al membrane ( « 150/um thick) that fulfills •REPRESENTING T H E BTEV PIXEL GROUP

178

179

Figure 1. Schematic view of a pixel half plane. Clearly visible are the shingled support structure with embedded cooling tubes and the hybrid pixel detector modules. The top layer is the high density cable routing signals and bias voltages.

also the purpose of shielding the sensor and associated electronics from RF pick-up. It encompasses 30 tracking stations, uniformly distributed over 1.28 m. The telescope length is matched to the expected length of the luminous region (az « 30 cm). Each station is composed of two pixel planes. Each plane has about 500,000 pixels in a 10 cm x 10 cm area, excluding a small central hole in the beam region. Fig. 1 shows one of the two L-shaped structures that make up a detector plane. These structures are retractable, as we are planning to move them away from the beam during injection. The design of a system satisfying these requirements without degrading the intrinsic hit resolution achievable is very challenging and the key aspects of our approach will be discussed below. The pixel devices are bump bonded to the readout electronics being developed at Fermilab. The hybrid sensor-readout assembly will be hosted on "shingled" carbon composite structures that allow close to 100% coverage over the active area. These structures will include integrated cooling channels. The unit pixel cell is 50 /mm x 400 jim. The technology chosen is n+/n/p+, namely the readout elements are located on the ohmic side of the device. This insures that the device is still operational after type inversion, even if it needs to be biased below full depletion voltage. In this approach the key design issue is the technology adopted to insure that adequate inter-pixel insulation

180

is achieved throughout the sensor lifetime. More details on our strategy will be given below. The final version of the electronics will feature analog readout including a 3 bit flash ADC in each pixel cell. The technology chosen to achieve the desired radiation hardness is 0.25 fim. CMOS process, implemented by two commercial foundries. Prototypes featuring a few cells have demonstrated that even after being exposed to 32 MRad of radiation, the analog performance of this front end still satisfies the experiment requirements. The readout electronics is close to the beam. Its control signal and output data are routed to the periphery through two flex circuits, a high density one, with very minute feature sizes, carrying the signals to the periphery of the pixel plane, connected to more conventional flex cables that fan the signals out of the vacuum box. The hybrid pixel detector is attached to a low mass support structure. The hybrid detector is going to be attached to the substrate at room temperature and then maintained at a temperature of about « —5°C to — 10° C. The substrate includes integrated cooling channels. Prototyping work is under way in collaboration with ESLI Technologies. The total material in a detector plane is 0.9% of a radiation length, mostly consisting of the silicon in the sensor and readout chip. 3

The hybrid silicon pixel detector

This paper will focus on our recent results on the sensor development and the bump bonding R&D 3.1

The pixel sensor

We have performed extensive test bench studies on sensors with p-stop interpixel insulation, which were manufactured for us by SINTEF, Oslo. Sensors were irradiated at the Indiana University Cyclotron Facility. Fig. 2 shows an example of the current-voltage data measured from a p-stop sensor including 12x92 cells and 10 guard rings. The measurement is performed by applying the reverse bias to the backplane and grounding all the pixel cells. We studied both oxygenated and non-oxygenated wafers. We found no significant difference between these two kind of wafers. In most of our measurements the breakdown voltage exceeded 400 V and did not change appreciably upon irradiation. The depletion voltage does not exhibit a steep increase as a function of the dose after type inversion. For example, at a fluence of 4 x 1014 p/cm 2 , sensors with an initial depletion voltage of 200 V showed a depletion voltage of

181

200

300

400

500

Voltage (V) Figure 2. Current-voltage characteristics for a SINTEF sensor before irradiation and after 4 x 10 1 4 p / c m 2 fiuence.

about 120 V. Irradiated and non-irradiated detectors will be studied with test beam data to be acquired in the summer of 2002. The static measurements indicate that the p-stop approach may be adequate for our application.

3.2

Bump bonding studies

We have identified two vendors for our initial study: Advanced Interconnect Technology, LTD (AIT) of Hong Kong, to explore the indium bump approach, and MCNC in North Carolina, USA, to explore the solder bump approach. To establish the reliability of each technology, as well as its radiation resilience, we have fabricated dummy detectors with rows of daisy chained bumps. The study of the electrical properties of these strips allows us to identify open or high resistance bumps. The resistance between neighboring strips allows to identify shorts. The indium bump detectors were on an even finer pitch than required (30 /xm). We monitored the quality of the bumps

182 Table 1. Rate of occurrence of problems (per bump) for dummy hybrid detectors manufactured with indium bumps (AIT) or solder bumps (MCNC). The cooling and heating cycles are described in the text. Problem bad contact after 1 yr bad contact after cooling bad contact after heating

In bumps 2.1 x 1 0 - 4 2.2 x 10~ 5 2.1 x 1 0 - 4

Solder bumps 4.0 x l O - 4 1.4 x l O " 4 6.3 x l O " 4

over the period of 1 year. We performed thermal cycling (alternating exposure to -10°C for 144 hours and +100° C for 48 hours in vacuum) and irradiated the dummy detectors with a 137 Cs source up to a dose of 13 MRad. Table 1 shows the rate of bump faults, manifesting themselves as high resistance developed through the daisy chained bumps. It can be seen that both techniques are highly reliable. Upon irradiation, the indium bumped detectors developed high resistance in a systematic pattern, almost in every first channel in a group of four. The grouping is a pattern characteristic of the specific sensor design. This effect is attributed to indium diffusion, with subsequent oxidation accelerated by radiation. Oxidized indium exhibited higher resistance. This effect was more prominent in the first cell of the group, as the edge cells have more exposure to Oxygen. Solder bumped hybrid detectors were affected mostly by a deterioration of the Al traces, that frequently became extensively flaky and bubbly upon irradiation. Thus the bump failure rates are extremely small in both technologies provided that we take enough precautions to avoid exposure to oxygen, as accelerated oxidation is an expected consequence of radiation exposure. 4

Acknowledgements

I would like to thank Gabriele Chiodini, Maria R. Coluccia, Simon Kwan and all the Pixel R&D group of Fermilab for useful discussion. I am particularly indebited with Marina Artuso for her precious suggestions in preparing this manuscript. References 1. A. Kulyavtsev, et al. BTeV proposal, Fermilab, May 2000. 2. J.A. Appel et al, hep-ex/0108014; to be published in Nucl. Instr. and Meth. A.

MONOLITHIC CMOS PIXELS FOR C H A R G E D PARTICLE TRACKING YU.GORNUSHKIN, G.DEPTUCH, M.WINTER IReS, IN2P3-CNRS/ULP, BP20, 67037 Strasbourg cedex, France E-mail: [email protected] G.CLAUS, W.DULINSKI LEPSI,

IN2P3/ULP, 23 rue du Loess, BP 67037 Strasbourg cedex 02, France

20,

The increasing need of high performance flavour tagging capabilities in particle physics experiments has triggered the development of a novel - fully integrated silicon pixel detector, called Monolithic Active Pixel Sensor. The first MAPS prototypes adapted to the detection of minimum ionising particles (mip) were designed and fabricated in standard CMOS technology. Their first tests demonstrate that the sensors detect mips with high signal-to-noise ratio, detection efficiency close to 100% and provide excellent spatial resolution. The main aspects of these results are summarised in this paper, together with preliminary results on the radiation hardness of MAPS. An outlook on the R&D started to adapt and integrate the sensors for future vertex detectors is also provided.

1

Principle of operation and detector description

Monolithic Active Pixel Sensors (MAPS) constitute a novel technique for silicon position sensitive detectors which allows integrating on the same substrate the detector element and the read-out electronics. The principle of operation of MAPS optimised for charged particle detection is based on a concept similar to that used in CMOS cameras - a rapidly growing up alternative to the CCD technology in digital photography and video applications 1. A thin epitaxial layer of low-resistivity silicon (doped to the level 1015 c m - 3 ) is used as a sensitive detector volume. The charge generated by the impinging particle at a rate of about 80 electron-hole pairs per 1 /xm is collected by a n-well/p-epi diode, created by n-well implantation into the epitaxial layer. The electrons liberated in epitaxial layer diffuse towards the diode within a typical time of a few tens of nanoseconds 2 . Because of the three orders of magnitude difference between the doping levels of the p-epitaxial layer and of the neighbouring p + + wells and substrate, potential barriers are created at the region boundaries, that act like mirrors for the excess electrons. Such a detector can be fabricated through a standard and therefore cost effective and easily available

183

184

_ ^

Pi Pi Pi =. Pi

Figure 1. Schematic diagram of MIMOSA read-out electronics.

CMOS process. In order to validate the ideas presented and to investigate the potential of the technology, four prototype chips with pixel arrays of slightly different design were fabricated. The first two prototypes, called MIMOSA" -1 and -2, were fabricated in two different CMOS processes: MIMOSA-1 in a 0.6-/xm process featuring an epitaxial layer of about 14 /on, MIMOSA-2 in a 0.35-/xm process with less than 5 /mi epitaxial layer thickness. MIMOSA-1 (resp. MIMOSA-2) contain four (resp. six) independent matrices of active elements having slightly different design. Each matrix consists of 64x64, 20 /xm wide, pixels. The individual pixel is comprised of three MOS transistors and a floating diffusion diode. The diode was implemented in the center of each pixel in most matrices. For one MIMOSA-1 matrix however, each pixel hosted four diodes connected in parallel, and one MIMOSA-2 matrix hosted two diodes per pixel. This was supposed to reduce the charge dispersion and collection time, at the expense of increased noise reflecting the higher node capacitance and the smaller charge-to-voltage conversion gain. Compared to MIMOSA-1, MIMOSA-2's design incorporates few new ideas, including features improving the read-out speed and the radiation hardness. The thinner epitaxial layer of MIMOSA-2 was expected to translate into less signal than MIMOSA-1. This feature was compensated by a lower noise level consecutive to smaller node capacitance and very small diode leakage current. Basic prototype parameters, such as the total conversion gain and the pixel equivalent noise charge (ENC), were determined with a 5.9 keV X-ray source of 55 Fe. Scematic design of MIMOSA chips is presented in Fig.l. More details on the chip architecture, principle of operation, results of the device simulation and of the tests with the X-ray source can be found in 2 ' 4 . "standing for Minimum Ionising M O S Active pixel sensor.

185

Figure 2. Left: Collected charge (most probable value) as a function of the cluster multiplicity. Right: Residue distribution of a track position measured by 1 diode/pixel MIMOSA-1 sensors (corrected centre-of-gravity method)

2

Test b e a m s t u d i e s

The first two MIMOSA chips were tested with a pion beam of 120 GeV/c at the CERN SPS. A beam telescope 3 of 8 planes of high precision silicon strip detectors, grouped in pairs of planes providing two orthogonal coordinates, was used to define the beam particle trajectory and intersection point with the MIMOSA plane with an accuracy of ~ 1 [im. The individual pixels were read out serially and an external 12-bit flash ADC unit digitised raw information from each pixel. The read-out clock cycle frequency used was 2.5 MHz for MIMOSA-1, and up to 10 MHz for MIMOSA-2. The off-line signal processing started with correlated double sampling (CDS) 5 to eliminate some dominant noise components. The remaining noise was evaluated from the first 250 events of every run and used to compute the pedestals subtracted from the CDS outcome as well as the noise entering the signal-to-noise (S/N) calculation. A cluster finding algorithm, using pixels with S/N > 5 as a seed, reconstructed clusters matching the beam telescope prediction. The charge collected in the cluster is displayed on Fig.2 (left) as a function of the cluster multiplicity. The values found are in agreement with the simulation results 6 . Typically, about 90% of the total charge was concentrated within a cluster of 3x3 pixels and the whole charge was contained within 18 pixels in most cases. As predicted, the 4 diodes/pixel option of MIMOSA-1 provided the most concentrated charge distribution. The charge collected with MIMOSA-2 (1 diode/pixel) was less spread than with MIMOSA-1, as expected from the thinner epitaxial layer. The coordinate of the track impact was measured using the charge distribution in the cluster. After a precise alignment of the prototypes w.r.t. the telescope, the v/idth of the distribution of the residue between the telescope extrapolation

186 Table 1. Major performances of the MIMOSA prototypes. Prototype Number of diodes/pixel Pixel noise (mean value) [e~ ENC] Signal-to-noise ratio (mean value) Detection Efficiency [%] Spatial resolution [/mi ]

MIMOSA-1 1 4 12 25 42 32 99.2 ± 0.2 99.5 ± 0.2 2.1 ± 0.1 1.4 ± 0.1

MIMOSA-2 1 9 22 98.5 ± 0.3 2.2 ± 0.1

and the MAPS reconstructed impact position varied from 1.7 to 2.5 /xm, depending on the chip design (Fig.2 right). The best resolution was achieved with MIMOSA-1, with 1 diode/pixel, the charge distribution being the widest and the S/N ratio being the highest. After subtraction of the track prediction resolution (~ 1 /xm), the intrinsic resolution of MIMOSA-1 came out to be better than 1.5 /xm. The performances of both prototypes are summarised in Table 1 and in 5 .

3

Preliminary results on radiation tolerance

The two prototypes were irradiated with 30 MeV/c protons and 10 keV Xrays. Irradiation effects were investigated by studying changes in the sensor response to 5.9 keV photons emitted by a 55Fe source. It was observed that 600 kRad X-rays provoke an increase of the diode leakage current by more than an order of magnitude. Similarly, a dose of 5 x 10 11 p/cm 2 induces a factor of 5 increase of the diode leakage current and a loss in the charge collected of about 40 %. On-going studies of sensors exposed to neutrons sources will clarify in which extend the observed sensitivity to protons is due to mechanical deformations of the epitaxy or if it originates from other degraded parts of the sensors (e.g. interfaces). Independently of the outcome of this study, present results on the radiation tolerance of CMOS sensors are encouraging, the first MIMOSA prototypes satisfying already the requirements of several applications (including those of a future e+e~ linear collider). Moreover, the influence of the temperature on the chip performances will be investigated in order to reduce its sensitivity and to recover (at least part of) the observed performance losses. Finally, forthcoming sensor prototypes will allow to continue exploring the radiation tolerance of the sensors and to design chips exploiting more and more efficiently the real potential of their technology.

187

4

Outlook: N e x t R & D steps

The validity of the CMOS sensor detection principle for charged particle being established, the present phase of development aims to explore the technology potential of the sensors (by testing alternative sensing devices, by studying and improving the sensor radiation tolerance, etc.) and to adapt them to various applications. Recently, a third prototype was fabricated in a standard 0.25 //m deep submicron IBM process with 8 x 8 fim2 wide pixels, and a fourth chip was manufactured in a 0.35 ^m AMS process with no epitaxy. These new arrays, which host various sensing devices, are currently being tested. Aiming to validate the sensors as building blocks of future vertex detectors, the design of the first real scale prototype (MIMOSA-5) was recently sent for fabrication in the same 0.6 /jm process as MIMOSA-1. The chips have the size of a CMOS reticle (i.e. about 19.4 x 17.4 mm 2 ) and consist of 4 independent matrices of 512x510 pixels with 17 fim pitch. They are stitched together along one direction across the wafer. Ladders of 5 or 7 chips are being achieved in this way. They will offer the first opportunity to test on real scale the performances observed with small scale prototypes, to investigate the chip production yield and the properties of stitching, and to experience the read-out of millions of pixels. A crucial issue of the next R&D steps will be the design of fast read-out micro-circuits integrated on the sensor substrate and including sparsification. 5

Conclusion

First prototypes of CMOS sensors designed for charged particle tracking show that this detection technique provides excellent performances (e.g. S/N ~20-=-40, detection efficiency >99%, spatial resolution down to ~ 1.5 /mi). Whether the latter can be reproduced on real scale will come out from the tests of the first large device, currently fabricated. The coming three years are expected to allow designing fast read-out micro-circuits with sparsification integrated on the chip as well as to explore extensively the real potential of this technology. The outcome of this programme will determine in which extent CMOS sensors can be used in future vertex detectors.

References 1. B.Diericks et al in Proc. IEEE CCD&AIS workshop, Brugge, Belgium (June 1997),p.Pl;

188

2. R.Turchetta et al., Nucl. Instrum. Methods A 458, 677 (2001); 3. C.Colledaniet al, Nucl. Instrum. Methods A 372, 379 (1996); 4. M.Winter et al., Proceedings of IEEE NSS Conference, October 2000, Lyon, France; 5. Yu.Gornushkin et al., Proceedings of the Vienna Conference on Instrumentation, February 2001, Vienna, Austria, and references therein; 6. G.Deptuch et al., Nucl. Instrum. Methods A 465, 92 (2001).

STATUS AND NEW LAYOUT OF THE ATLAS PIXEL DETECTOR P. NETCHAEVA INFN Genoa, Italy on behalf of the ATLAS Pixel collaboration [IJ

The ATLAS Fixe! detector is based on a set of radiation-hard electronics chips able to resist a dose of SOOkGy. The implementation of these chips in the DMILL technology did not give the expectedresults.Re-design of the radiation-haw! clips in DeepSubMicron technology is ongoing, but has implied a one and a half year delay in an already tight schedule. Major layout changes have therefore been necessary to allow installation of the ATLAS pixel detector at LHC start-up. This paper illustrates the status of the ATLAS pixel project, the motivations for the new layout, the way this should be implemented and the prototype fabrication and testing.

1. INTRODUCTION

The ATLAS Pixel detector has been already described elsewhere [1] and will be only briefly recalled here. This detector should be able to measure three space points in the pseudorapidity region up to |T|| = 2.5. It is made of two barrel layers, initially having 10.1 cm and 13.2 cm radius, plus so called b-layer of 4.3 cm radius. The main aim of the b-layer is to allow better impact parameter resolution for B-physics and for ©-tagging. The forward region is covered by set of 5+5 disks with internal and external radii of, respectively, 12.6 cm and 18.7 cm. The detector system is composed of modular units (fig.l).

Fig.l. ATLAS Pixel module.

Each module consists of a silicon sensor tile, sixteen front-end readout integrated circuits (FE chips) and a kapton flex hybrid. The flex hybrid distributes power and control signals to the FE chips and allows reading them out through a module control circuit (MCC). Passive components including termination resistors, decoupling capacitors and temperature sensor are also included. The sensitive area of a module, i.e. of a sensor tile, is 16.4 mm x 60.8 mm. The FE chips are connected to the pixel cells

189

19Q

through bump bonds. The size of one pixel is 50 am x 400 um and each FE chip serves 18 x 160 pixel cells. The flex hybrid is glued to the backside of the sensor tile; electrical connections from the MCC and the 16 FE chips to the flex hybrid are done through ultrasonic wedge bonding. 2. RADIATION HARDNESS

2.1 Sensors

Radiation damage in the severe LHC "environment can result in a voltage for full depletion of silicon detectors, which may exceed the maximum allowed operation voltage, and thus requires the detectors to be operated only in partial depletion [1]. The baseline design of ATLAS Pixel silicon sensors is characterized by: ® n + pixels on oxygenated n-bulk material (double-sided processing) to allow partially depleted operation [2], • moderated p-spray isolation to allow for high voltage breakdown after the type inversion, ® a bias grid to allow the sensor testing before module assembly. The sensors radiation hardness to 10 years of LHC operation has been proven [3]. One of the proofs recently obtained is the interpixel isolation test, which demonstrates no ionization-induced damage (fig.2). The sensors under test have been irradiated up to 500 kGy with 20 keV electrons; bias voltage of 500 V has been applied during the irradiation. The I-V curves show the pinch off at higher voltages for the irradiated sensor, but the interpixel isolation (about 100 MOhm at 100 V bias) is still sufficient.

upteSMkSy, with 500 V Was during Irffufatfon j

so

Fig.2. Interpixel isolation test

The preproduction of 150 good tiles has been done; careful tests have confirmed that the ATLAS specifications are met. The production will start in January 2002.

191

2.2 Electronics The four integrated circuits: FE chip (16 per module), MCC (event builder for 16 FE chips), VDC (VCSEL Driver Chip to drive data off-detector) and DORIC (to decode/encode clock and control signals) have been implemented in the DMILL radiation hard technology. Despite the large efforts of the Pixel collaboration this technology has not been able to produce the most complex chips (FE and MCC) with acceptable yield. Therefore the DMILL design work was stopped and maximum priority had been given to translate the various designs to the DeepSubMicron (DSM) technology. First digital DSM test chip had some limited functionality but worked roughly as expected. The analog test chip had been submitted to IBM and TSMC in February / March 2001. This chip contained the preliminary designs of analog blocks of 20 pixels and other analog circuits, as well as the final layout of the critical items. On-line results and first tests done after irradiation (at a dose of 610 kGy with 55 MeV protons) indicated little or no change in performance of the test chip. 3. INSERTABLE LAYOUT The failure of the DMILL design and the transition from DMILL to DSM has generated 1.5 years delay in the Pixel detector schedule and made impossible the "ready for installation" date initially foreseen (april 2004). This date was required to install the Pixel detector together with the Barrel Inner detector. The "Insertable" layout decouples the Pixel detector from the rest of the Inner detector and allows its independent installation later. This decoupling is obtained installing a 7 m long support tube together with the Barrel Inner detector. The Pixel system can then be slided in this support tube when the vacuum is broken and the forward section of beam pipe is out. The central section of beam pipe is integrated in the Pixel system and move with it. The Pixel system, services and beam pipe are prepared on surface, lowered using a temporary support and rolled into the support tube held by SCT barrel. The services for all but blayer should go out both sides. The b-layer services go out on one side to allow installation and dismounting together with the beam pipe in place. To fit the SCT bore the Pixel system had to be squeezed (the outer barrel radius has been changed from 14.2 cm to 12.2 cm) and shortened (fig.3). The number of disks is reduced (from 5+5 to 3+3); all disks have been made equal (8 sector each) and come closer to the interaction point. The inner radius of the Pixel system has increased from 4.1 cm to 5.0 cm because of the beam pipe radius increase. In total, the detector active area is reduced by 17 %. The Insertable layout has some reduced performance. In the plot for probability to have less than 3 hits (fig.4) we see some losses due to clearances in the barrel section.

192

Side C

111111 ifl 111111 M i l l Hit" HI H\

.—



2 -f

Side A

* and a bulk resistivity between from 1.4 and 7kficm. Some sensors were produced using the oxygenation (OX) technique which allows a reduction of the operating voltage for sensors irradiated with charged particles 2 . The oxygenation was performed from a local oxygen layer grown into the bulk at 1200° C during 100 hours, resulting in an oxygen concentration of about 3 • 10 17 cm~ 3 . Selected detectors were pre-irradiated by 25 . . . 34 MeV protons to an equivalent fluence ofl...2-1014n(lMeV)/cm2.

226

2.2

Readout chips

Two successive generations of the APV chip were used to read out the sensors:!) APV6 manifactured in a 1.2//m radiation hard CMOS process and 2) APV25 made in commercial 0.25 pm deep sub-micron CMOS, which provides intrinsic radiation tolerance 4 . The 128 channels APV chip consists of a preamplifier followed by a CRRC shaper with a time constant of r = 50 ns. The shaper output is sampled at 40 MHz and stored in an analog pipeline of 192(APV25) cells. When a trigger arrives, three consecutive samples are processed by a switched capacitor filter which performs a deconvolution algorithm that narrows each pulse down to one single clock cycle in order to identify the exact bunch crossing. The predominant noise source in the detector and readout system is the preamplifier input transistor of the preamplifier. The measured noise in deconvolution mode for APV6 is ENC=1000 + 46 p F - 1 and 400 + 60 p F - 1 for APV25. Other noise contributions such as leakage current, strip resistance or bias resistor together account for about 430 e - which is quadratically added 1. With a capacitive load of 16 pF, which is typical for CMS detectors, we expect ENC = 1426 e~ with the APV25 in deconvolution mode, which is in good agreement with the measured value of 1430e~. One to three readout chips (see Table.l) were mounted on hybrids assembled with a pitch adapter on the detector frame carrying two daisy-chained sensors. Thus the total strip length was 12 cm (16cm for VB25).

3

Experimental setup

All detector modules under study were housed in a cooling box which was operated by two water cooled peltier elements with a total cooling power of ~ 200 W at A T ~ 40° C (Thox = - 2 0 ° C). The box was flushed with dry nitrogen to prevent water condensation. The tests were performed in the PSI 7rMl beam line which provides 350MeV/c pions or protons with a rate of up to 9 kHz/mm 2 and a beam spot of approximately 50 x 50 mm 2 FWHM. In the present study we operated the detectors mostly with pions. Although the 350 MeV/c pions are approximately minimum ionizing particles (MIPS), secondary reactions can produce heavily ionizing particles (HIPS) with up to 1000 times larger ionization losses.

227

4 4-1

Results Front-end electronics

Irradiation may cause two kinds of damages to the readout electronics: permanent or transient. Due to its deep sub-micron fabrication process, the APV25 should be intrinsically radiation tolerant. Eight APV25S1 chips were irradiated to 1.87 • 10147r/cm2 at 300MeV/c momentum 5 . No critical, irreversible damage was observed. The irradiation did not affect the calibration signal SNR within ±5%. We have observed a dependance of the calibration SNR on the temperature of about ~ 25% for AT=30°C. This dependance was expected and is due to variation of the chip settings which have to be optimized for each temperature. The charge released by HIPS can result in the nipping of an APV register cell, called a Single Event Upset (SEU). We measured a total cross section of approximately 2 • 10~ 12 cm 2 for such SEU, corresponding to 2 • 10~ 15cm2 for a single flip-flop cell. Single event upsets also occur in the analog circuitry, but are self-repairing and appear as a negligible increase in noise background. 4-2

Detectors behavior

The Signal to Noise denned as „ " ' ' " ' " • , where Noiseciuster=y *^ , . '-, was analyzed from runs collected in the 350 MeV/c pion beam, at a rate below 100Hz/mm 2 . Figure 1 left presents the most probable (MP) SNRcius values as a function of Vbias- The averaged rms noise at Tb ox = —10° C is independent from the Vuas within 5%. The highest SNR of about 16.5 was achieved with the non-irradiated PD26 detector. As expected, the APV25 readout provides an ~30% increase of the SNR due to lower noise compared to the APV6. The significant difference in the signal obtained for the irradiated detectors BA2 and PD27 can be explained by different production technologies. The BA2 was produced with < 111 > crystal orientation, while the PD27 was manufactured from < 100 > silicon using the oxygenation technique 6 . The irradiated oxygenated sensor (PD27) demonstrated almost the same SNR as the non-irradiated (PD26) at Vbias = 550 V while the non-oxygenated irradiated sensor (BA2) had lost ~ 20% of the signal in comparison with the non-irradiated one (BAl). For the irradiated sensors the maximum SNR was achieved at Vbias ~ 3 times above the expected Vd ep i et j on . Note that all irradiated detector modules have shown stable operation at a bias voltage of 550 V.

228 •

' •

*****

" . : - . • "

m

» ."1 •

A

* A * * T m * 200

BAImirr(atT=I8C) PD26nirr. BA2irr. PD27iir. VB25nirr. PDImirr. HWfrr.

X

•

•

m

1 • A

A

* •

** • m• •

A

A

A

„ A PD26mrr. M BA2irr.

•

PD27UT.

300

Figure 1. Left is a SNR in deconvolution mode versus Vbias at T=—10° C (except BA1). On the right is a cluster size versus Vbias m deconvolution mode.

The cluster size (figure 1 right) is larger for irradiated detectors presumably due to an under-depleted zone near the p + strips and charge trapping. The dependence of the cluster size on Vbias follows the dependence of the interstrip capacitance Cj n t . For irradiated sensors, Ci nt decreases with Vbias, especially for the < 111 > crystals, while for non-irradiated sensors, Ci„t remains constant and for irradiated < 100 > crystals, the dependence on Vbias is weaker. Due to the absence of precise independent tracking and large multiple scattering we could not measure the absolute efficiency. However we can roughly estimate the knee of the efficiency plateau by using the detectors under test as a tracker, see figure 2 left. To study the SNR uniformity the beam spot has been moved along the strips. No variation in the SNR has been found for the largest VB25 detector in a beam spot scan from one end to the opposite end. The measured variation of the SNR across the sensor was below 2.5% for all detectors. Unlike our measurements at PSI, where the beam incidence was usually perpendicular to the detector plane, a wide-spread angular distribution is expected in CMS l. An angular scan has been performed with the 140 ^m pitch VB25 module at room temperature. In figure 2 the maximum probability signals and the cluster sizes are presented versus the incident angle a to

229 APV2S angla Man - signal dustar width Decomokition Mods.Standard MtHng*. 360 MeV/c protoru and pkxil • A 0.9S

Pretoria- duster Width (measured) Pretoria - Fit sqrtfconsP+tan9) F ^ s - C l i i s t e f Width (measured) Piona - Fit aqrt(conet*+tan*)

S .-' _.-•''

. - - * • " * * " %

0.9

-

0.8S

-

/ a

•

20

so

«

so

eo

Taanglean

».»

APV25 angla scan - duatar algnal i. Standard MltkigK 350 MeV/c prolont and plant

0.7S

-

0.7

-

•

BAlirrad.

Pretoria - MP SNR (meaaurad) • Protona - Fit 1/coa Piona - MP SNR (meaaurad) Rons - Fit 1/cos

*,/' t,'"'

•"a^"'"

0.65

1 , . . 1

0

2

4

t

8

10

12 14 SNK

10

20

30

40

50

60

Tlangbnll

Figure 2. Efficiency versus MP SNR for the BA2 detector operated at T=—10° C in deconvolution mode on the left. On the right cluster size (top) and most probable signal (bottom) of the VB25 detector module versus incident angle for 350 MeV/c pion and proton beams in deconvolution

the detector plane for pion and proton beams at a momentum of 350MeV/c. The signal dependence is described with l/cos(a) and the cluster size, with y'c 2 + tan2 [a) functions, where c denotes the cluster width at perpendicular incidence. The ratio of SNRP is about 6.6 and thus in good agreement with the calculation from the restricted Bethe-Bloch theory which predicts 6.0. The detectors have demonstarted a stable operation in the high intensity pion beam at 9 kHz/mm 2 , the leakage current have increased with fluence Ileak ~ $OL, a — 8 1 0 ~ 1 7 A / c m .

5

Summary

Different type of silicon strip detectors were tested in a hadron beam under conditions close to what is expected at the LHC. With a strip length of 12 cm and irradiated oxygenated sensors, we have obtained signal-to-noise value of 15.5 for the APV25 and 10 for the APV6 readout chips in deconvolution mode. The signal-to-noise is uniform along and across the strips within a

230

level of 2.5%. It has been shown that the sensors and the readout chips do survive in the harsh radiation environment of LHC. No critical damage could be observed on the readout chips, and the single event upset rates are sufficiently low so they will cause only negligible corruption of data. In an angle scan, the detector modules behaved as expected from geometrical relations and the measured signals were consistent with the restricted Bethe-Bloch theory for pions and protons. 6

Acknowledgments

We would like to thank K.GabathuIer and D.Renker for their help at PSI and L.Shektman for sharing the beamline with us. References 1. CMS Tracker Technical Design Report, CERN/LHCC 98-6, 1998. 2. A.Ruzin for RD48, NIM A 447 (2000), 116. 3. T.Beckers et al, Proceedings of 9-th Vienna conference on instrumentation 2001(to be published in NIM A). 4. M.French, APV User Manuals, h t t p : //www. t e . r l . a c . uk/med. 5. M.Friedl et al, Proceedings of the Vertex 2001 conference (to be published in NIM A). 6. N.Demaria et al, NIMA 447 (2000) 142-150.

STATUS OF T H E CMS PIXEL D E T E C T O R T. ROHE Paul Schemer Institut, 5232 Villigen, Switzerland e-mail: [email protected] for the CMS Pixel Collaboration The innermost layers of the CMS tracking system will consist of pixel detectors. They will allow pattern recognition in the high track density and will be used as vertex detector. An overview of the system and a status report on the different components will be presented. Emphasis will be given to the latest developments in 2001: The first submission of a full-size radiation-hard readout chip and the latest sensor prototyping.

1

Detector Layout

The tracking unit of the CMS experiment at the Large Hadron Collider (LHC) will contain hybrid silicon pixel detectors for track reconstruction and btagging 1 . It will consist of three barrel layers and two end disks at each side. The barrels will be 53 cm long and placed at radii of 4.4 cm, 7.3 cm, and 10.2 cm (fig. 1). They cover an area of about 0.8 m 2 with roughly 800 modules. The end disks are located at a mean distance to the interaction point of 34.5 cm and 46.5 cm. The area of the 96 turbine blade shaped modules in the disks sums up to about 0.28 m 2 . In the first years when LHC has not reached its final luminosity only the two innermost barrel layers and the first disk on each side will be installed. This system will represent about half of the final system and provide a two hit coverage up to a pseudorapidity" of \rj\ < 2.1. By adding the 3 r d barrel and the 2 n d disk the system will provide three hits over about the same solid angle without extending the region of two hit coverage. In order to achieve the best vertex position measurement the spatial resolution of the sensor should be as good in the ^-direction (parallel to the beam line) as in (r, ip) and therefore a squared pixel shape with a pitch of 150 x 150 um2 was adopted. To improve the spatial resolution analog interpolation between neighbouring channels will be performed. The strong Lorentz deflection in the (r, (^-direction caused by CMS' 4 T magnetic field is utilized to distribute the signal onto several channels. Hence the detectors are not tilted in the barrel layers. The resolution along the z-axis is determined by a

n = — lntan(0/2) where 0 is the track angle relative to the beam axis.

231

232

Figure 1. Perspective view of the CMS pixel system.

the pixel pitch in the region with low pseudorapidity and by charge sharing if the tracks hit the sensors under an angle where the typical cluster size can exceed values of 6 or 7. The best resolution will be reached at the point where the charge is distributed over about two pixels. In the disks where the charge carrier drift is hardly affected by the magnetic field the modules are tilted about 20° resulting in a turbine like geometry visible in fig. 1. 2

Modules and Mechanics

A picture of a barrel module is shown in fig. 2. A 66.3 mm long, 18.45 mm wide, and 300, nm thick sensor is bump bonded to 2 x 8 readout chips. The bump bonding procedure using Indium has been developed at PSI. It is currently used to assemble 64 pixel modules for an experiment at the Swiss synchrotron light source (SLS) at PSI 2 ' 3 . This ensemble is glued with the chips down to a 270 urn thick silicon base plate which is attached to the cooling frame with small screws. Silicon is chosen as material to avoid mechanical stress

233

bumps

readout chips

silicon base plate

Figure 2. View of a barrel module. In the cross section the vertical scale is raised by a factor of 5.

due to thermal expansion. A 50 um thick polyemide hybrid that is thermally matched to silicon is glued on the sensor's back side. The electrical connection to the readout chip is done via wire bonding. The hybrid is equipped with passive components and the so called token bit manager chip managing the readout of the system. Clock, control and data signals are transferred to the barrel periphery via a copper-on-Kapton cable glued and wire bonded to the hybrid. The power is brought in by extra aluminum wires. At both ends of the barrel an end flange is situated where the cables are grouped into sectors and brought to the end of the tracker volume via a 2.2 m long service tube. On this service tube the conversion from electrical to optical signals and vice versa is performed. The end cap modules look slightly different due to their trapezoidal shape. The width of the sensors varies from 2 readout chips per sensor in the innermost region to 5 at the outer end of each blade. The hybrid is placed between the readout chip and the base plate *. The full pixel detector including the service tubes can be preassembled and inserted into CMS as the last component. It will have to be removed at least every second year of LHC running for beam pipe bake out and replacement of the innermost layers which suffer most from radiation damage.

234

Figure 3. Readout scheme of the CMS pixel readout chip. Data of hit pixels will be transmitted to the column periphery via a fast scan and stored there until either rejected or read out via the time stamp and data bus. The control & interface block contains a logic to program the chip and set some reference voltages.

3

Detector Readout

The high bunch crossing rate of 40 MHz requires LHC experiments to readout while data taking continues. This involves a complicated scheme of buffering the data in frontend pipelines up to the time when the trigger decision arrives. The readout architecture of the CMS pixel detector is explained in 4>1. 3.1

Chip Architecture

The readout chip as the most crucial component of the readout chain uses a column drain architecture as described in 5 ' 1 . This architecture assigns as many tasks as possible to the column periphery located at the edge of the chip as shown in fig. 3. This approach keeps the pixel unit cell simple. In its present implementation it contains only about 125 transistors, much less than those of other LHC experiments.

235

When a pixel is hit it notifies the column periphery via a fast OR. A time stamp is created instantly and a token search mechanism is initiated scanning all pixels of a double column as indicated in fig. 3. If a hit cell is identified, its address and analog signal is transmitted to the data buffer in the periphery. In contrast to a fixed association of data buffer cells with each time stamp, a scheme using a pool buffer for all pixel cells of a double column was adopted. It allows to store a variable number of pixel hits per time stamp and ensures that large variations in the hit multiplicity are accepted. This is important for events where the pixels are inside a high pt jet or in case of heavy ion collisions. The readout of several chips in a module is controlled by the token bit manager chip. It sends a token flag to a group of daisy chained readout chips scanning all chips for hit double columns in a similar way as done for the pixel cells in one double column. When a hit double column is found data is sent to the DAQ system via the time stamp & readout bus as indicated in fig. 3. In addition the readout chip will contain a control & interface block for chip programming and setting of reference voltages. 3.2

Latest Results from Prototyping

A chip (PSI 41) with the architecture described above has been produced in the radiation hard DMILL process. It contains 36 x 40 pixel cells which is about half the number of the final chip with 52 x 53 pixels and totals roughly 240000 transistors. The pitch is 150 x 150 um 2 , according to the technical design report 1. The following features are implemented: • Final column drain architecture as explained above. In addition a double hit capability is implemented allowing to record an additional hit in the double column during a hit scan. • The complete double column periphery is included with 8 time stamp buffers and 24 pixel data buffers. • Fully functional readout chain. The readout takes 6 clock cycles per hit. The column and pixel addresses are coded in five analog levels. The pulse height is also read out analog. • Some variations of the analog block are implemented for final optimisation. Still missing in this prototype is the control & interface block which contains 21 DACs to set reference voltages, voltage regulators for the supplies, and a fast I2C-interface for chip programming. It is included in the final chip (PSI 43) submitted in autumn 2001.

236 Rgpr

Rgsh

I—wv-

i—y^r-

Comp

(a)

cnH>-

250fF

1.7fF

Preamplifier

—ycV—i

(b)

A

Shaper

X

Sample and hold

i—yw-

Hh

OT^> Figure 4. Two realisations of the analog block of a pixel unit cell: (a) with a source follower (b) alternative option with lower power consumption.

Tests performed with the PSI 41 chip were successful. Pixels can be enabled or masked, thresholds can be set and tuned individually for each pixel, and calibration pulses can be injected. The column drain mechanism performs as expected. A speed of 2 GHz was reached in the hit scans. Time stamp and data buffers and the event assembly is working well. Problems showed up in the readout and clocking speed which is limited to 15 MHz and 35 MHz respectively instead of the required 40 MHz. However the reasons of the speed limiting problems were identified and fixed for the next chip generation PSI 43. The power consumption of the chip is in the order of 90 uW per pixel equally shared by the digital and the analog part. A significant fraction of the analog power is consumed by the source followers (fig. 4a), which are fed by a separate supply voltage. In order to simplify the system, replacement options for the source followers were investigated. While the first one between preamplifier and shaper can be omitted by slightly reducing the feed back capacitance, the second one has to be replaced as shown in fig. 4b. The power needed by the current mirror is much decreased compared to the source

237

followers and the voltage adjustment is less critical. Therefore this supply voltage can be derived from the analog power line, reducing the number of required supply voltages from four to three. The performance of this circuit is superiour to the original one. The risetime at the end of the shaper which determines the time walk of the comparator is still fast and was measured to about 20 ns with an artificial input load capacity of 106 fF. The output to the sample-and-hold also displays a fast peaking time but a slower return to baseline due to the p-MOS transistors. This provides further robustness as timing of the sampling is less critical. A further advantage of the new circuit is the reduction of the power consumption by 15-20%.

3.3

Future Plans

The full size radiation hard chip PSI 43 with all features necessary for an operation in CMS including the control and interface block is currently in production using the DMILL process. The recticle will also contain the token bit manager chip. Delivery is expected in spring 2002. A translation of the chip into a radiation hard 0.25 um-technology is planned. This technology offers the possibility of a further pitch reduction with a cell size of 100 x 150 um 2 instead of the previous 150 x 150 urn2. In the disks the reduced pixel dimension measures the r-direction with improved accuracy. In the (^-direction the spatial resolution is favoured by charge sharing induced by the 20° tilt of the modules. In the barrel the zresolution in the non-central region is determined by charge sharing due to the tilt of the tracks. A pitch reduction along z would only improve the resolution were the hit multiplicity is well below two, which is the case for -q < 0.5. For the barrel the pitch reduction is realised along the (r, 12. The implemented guard ring structure consisting on 10 floating rings with increasing distance towards the edge displays a very good performance up to voltages of l k V 9 . In order to continue the sensor developement and to further optimise the design, a second sensor prototype has been submitted in 2001 10 . It contains several p-stop design options with the aim to reduce the bias dependence of ''fluence is normalised to lMeV neutron equivalent n e q /cm 2 .

239 the interpixel resistance. This will be reached by wider p-stop openings and larger gaps between the p-stops themselves. This implies that only one p-stop atoll ring per pixel will be used. To improve the post radiation breakdown behaviour two aproaches will be followed. One possibility are field plates covering the lateral pn-juction of the p-stops which are held on p-stop potential. This method used in power electronics since the late 1960s 13 showed to be quite successful 12>14. The other approach addresses directly the root of the problem, i.e. the reduction of the p-stop dose from currently about 5 x 10 13 c m - 2 to the minimum value possible. This automatically leads to the p-spray technique 15 which is known for its good high voltage capability in the irradiated state 16>17. A final decision on the sensor design will be made after the inverstigation of the second prototype submission in 2002. 5

Summary

The CMS pixel detector has been described. Latest developement is the submission of a full size radiation hard readout chip in DMILL technology satisfying the CMS specifications. All critical features have been investigated in previous readout chips with a reduced size. A translation of the chip into a radiation hard deep sub-micron CMOS design is planned. This allows a moderate decrease of the pixel size. Sensors have been produced and successfully tested in 2000. A second prototye for a further optimisation has been submitted in 2001. References 1. The CMS Collaboration, CMS Tracker Technical Design Report, CERN/LHCC 98-6. 2. Chr. Bronnimann et al., A pixel read-out chip for the PILATUS project, N I M A465 (2001) 235-239. 3. E. F. Eikenberry et al., PILATUS: A 2-D pixel detector for protein crystallography, Presented at the 10 th International Workshop on Vertex Detectors "Vertex 2001", September 23 rd -28 th , 2001 in Brunnen, Switzerland, to be published in NIM A. 4. D.Kotlinski et al., The CMS pixel detector, N I M A465 (2001) 46-50. 5. R. Baur et al., Readout architecture of the CMS pixel detector, N I M A465 (2001) 159-165. 6. B.Hendrich and R. Kaufmann Lorentz angle in irradiated silicon, Presented at the 5 t h Position Sensitive Detector Conference, London, 1999, accepted for publication in NIM A. Accessible via

240

http://cms.web.psi.ch/cms_conference.reports.html 7. M. Aleppo et al., A measurement of Lorentz angle of radiation-hard pixel sensors, N I M A465 (2001) 108-111. 8. C. Troncon et al., A measurement of Lorentz angle and spatial resolution of radiation-hard pixel sensors, accepted for publication in NIM A 9. R. Kaufmann, Developement of Radiation hard Pixel Sensors for the CMS Experiment, PhD thesis at the faculty of mathematics and science at the university of Zurich, Switzerland, 2001. 10. T. Rohe et al., Sensor development for the CMS pixel detector, Presented at the 5 t h International Conference on Large Scale Applications and Radiation Hardness of Semiconductor Detectors, July 4 t h -6 t h , 2001 in Firenze, Italy, accepted for publication in NIM A. 11. D.Robinson et al., Noise studies of n-strip on n-bulk silicon microstrip detectors using fast binary readout electronics after irradiation to 3 x 1 0 1 4 p c m - 2 , N I M A426 (1999) 28-33. 12. Y. Unno et al., Novel p-stop structure in the n-side of silicon microstrip detector, presented at the Hiroshima symposium on semiconductor devices, held 1997 in Mebourne, Australia. Submitted to the conference proceedings (not published). Accessible via http://jsdhpl.kek.jp/~unno/notes.html 13. B. J.Baliga Modern power devices, Wiley, New York, 1987, pp 116 and references therein. 14. T. Nakayama et al., Radiation damage studies of silicon micro strip sensors, IEEE Trans. Nucl. Sci. Vol. 47, No. 6, December 2000, p 18851891. 15. R. H. Richter et al., Strip detector design for ATLAS and HERA-B using two-dimensional device simulation, N I M A377 (1996) 412-421. 16. M.S. Alam et al. The ATLAS silicon pixel sensors, N I M A456 (2001) 217-232. 17. T. Rohe et al., Design and test of pixel sensors for the ATLAS pixel detector, N I M A (1999) 55-66.

F A B R I C A T I O N OF M I C R O S T R I P D E T E C T O R S A N D INTEGRATED ELECTRONICS ON HIGH RESISTIVITY SILICON

G.-F. DALLA BETTA* M. BOSCARDIN, P. GREGORI, N. ZORZI ITC-irst, Divisione Microsisttmi, 38050 Povo (TN), Italy G. BATIGNANI, S. BETTARINI, M. CARPINELLI, F. FORTI, M. GIORGI, A. LUSIANI, M. RAMA, F. SANDRELLI, G. SIMI INFN-Pisa and Universita di Pisa, 56010 S. Piero a Grado (PI), Italy

INFN-Trieste

L. BOSISIO, S. DITTONGO and Universita di Trieste, 34128 Trieste, Italy

G. U. PIGNATEL Universita di Trento, 38050 Mesiano (TN), Italy P. F. MANFREDI, M. MANGHISONI, L. RATTI, V. SPEZIALI, G. TRAVERSI INFN-Pavia and Universita di Pavia, 27100 Pavia, Italy V. RE Universita di Bergamo, 24044 Dalmine (BG), Italy A fabrication technology has been developed at ITC-irst (Trento, Italy) for the realisation of silicon microstrip detectors with integrated front-end electronics, to be used in high-energy physics and space experiments as well as in medical/industrial imaging applications. The main technological issues are addressed, and experimental results from the electrical characterisation of the first prototype batch are reported, showing that good quality transistors are obtained within the proposed technology while preserving the basic detector parameters.

1

Introduction

The possibility to integrate at least part of the front-end electronics on the same silicon radiation detector substrate can greatly simplify the mechanical assembly of the read-out system. Owing to the minimisation of the stray capacitance associated with the detector-preamplifier connection, this approach can also enhance the noise performance for detectors having a low output capacitance, such as drift chambers and pixel detectors, whereas for microstrip -TEL.+39-0461-314543, FAX. +39-0461-302040, E-MAIL: DALLABESITC.IT

241

242

detectors, which feature a relatively high capacitance ( « l p F / c m ) , the advantages of integrated electronics are less evident. Nonetheless, a fully integrated system is very appealing in applications where compactness, weight, amount of material are crucial. For instance, an integrated preamplifier at the end of a strip detector would allow to move the rest of the readout electronics, which in traditional systems is right next to the silicon detector, further apart, significantly reducing the amount of material and complexity in the active detection area. Besides, a very dense stacking of detectors would become possible, with applications in X-ray detection or active targets. Finally, integrating the entire readout chain at the end of the strip, an extremely compact system with very few connections and external components would be obtained, which is ideal for space applications. To this purpose, we have modified the fabrication technology developed at IRST for PIN detectors with integrated N-JFET's 1>2, in order to realise microstrip detectors and integrated read-out electronics on high resistivity silicon. We report on the main technological issues and on selected results from the electrical characterisation of the first prototype batch. 2

Design of detectors and integrated electronics

As a first step toward the realisation of a fully integrated detection system, we have considered a monolithic structure consisting of a microstrip detector with integrated N-JFET (triode) in the source-follower configuration. Fig. 1 shows the schematic diagram of the monolithic structure, consisting of a strip detector (SD), either (A) DC- or (B) AC-coupled to a source follower, to be connected via the integrated capacitance CA to an external readout circuit (charge sensitive amplifier plus shaper). Theoretical analyses and circuit sim-

Figure 1. Schematic diagram of a microstrip detector with (A) DC-coupled, (B) ACcoupled integrated N-JFET in the source-follower configuration.

243

illations have been carried out, assuming the external electronics to be implemented with the analog section of the AToM chip, developed for the readout of BaBar detectors 3 . Good electrical figures have been predicted, also in terms of Equivalent Noise Charge (ENC), the value of which was found to be in the order of 500 e~ rms in the shaping time range from 100 to 400 ns 4,B . 3

Device fabrication

IRST process for PIN diodes and integrated N-JFET's, detailed in l, has been further developed to include polysilicon (low and high resistivity) and recessed coupling capacitors with a stacked-dielectric insulator (SiC^-TEQS), while maintaning the same basic approach which features: (i) p + and n + implants (shallow and deep) and thermal diffusion for the transistor realisation and (ii) back-side, P-doped poly-Si gettering to ensure low diode leakage current. Moreover, the process thermal budget has not been altered, so as to preserve the most critical characteristic, i.e., the JFET doping profile in the gateregion. The schematic cross section of a monolithic structure, consisting of a microstrip detector and a front-end JFET, is shown in Fig. 2.

I n' SI substrate

I

Figure 2. Schematic cross-section of a monolithic strip-)-JFET structure (not to scale).

4

Experimental results

Measurements on test structures evidenced an adequate process control of the main parameters, such as: (i) diode leakage current (~ 0.5 ± Q.lnA/cm2 at M l depletion); (ii) polysilicon sheet resistance, both for high-resistivity (14.1±0.6 kO/sq.) and low-resistivity (402±8 fi/sq.) resistors; (iii) dielectric thickness of the insulators employed in poly/p + (212±3 mm) and poly/metal (193±7 urn) integrated capacitors. JFET's were tested on-wafer by means of an automatic probe-station. All measurements were carried out at substrate reverse voltage, Vs„&=6GV, higher

244

than the wafer depletion voltage. As an example, Fig. 3 shows the transfer characteristic of a transistor having an aspect ratio W/L=1000/xm/4/zm, namely the same device adopted in the structures of Fig. 1. The pinch-ofF voltage is about —1.15V, while the drain saturation current, Id ss , is about 3mA, resulting in a high transconductance, gm ~7.5mS. The output characteristics of the same device at different Vgs values are shown in the inset of Fig. 3: a good saturating behaviour in the pinch-off region is observed, and high output resistance (rout) values were measured. The main electri-

gate-to-source voltage, V ^ (V)

Figure 3. Id-V 93 characteristic of a J F E T with W/L=1000/mi/4/mi, with output characteristics at different Vgs values in the inset.

frequency (Hz)

Figure 4. Spectral density of the noise voltage as a function of frequency for three J F E T samples with W/L=1000/im/4pm.

cal parameters of JFET's having different width (W) with the same length (L=4/im) are reported in Table 1, evidencing a correct scaling of the electrical figures with the device width. Note that in the Vgs and Vds range of practical interest, the JFET gate current, I 9 , is low, its value being dominated by the leakage current of the p-well/n-substrate junction. On the contrary, the values measured for the input capacitance, C s s s , are quite high, particularly for the device with W=1000/im. Thus, in order for the ENC performance of the strip-)-JFET structure not to be degraded, a smaller transistor width should be preferably used, allowing for a better g m vs. Cgss trade-off. Noise tests have also been performed on JFET's: as an example, Fig. 4 shows the spectral density of the noise voltage as a function of frequency for three devices having W/L=1000)um/4/zm, biased with Iat < 0.1) between the orthogonal coordinates of the 2D projective readout improves the reconstruction capability for multiple hits. High rate tolerance and low discharge probability make the GEM detectors well suited for operation in intense muon and hadron beams.

1

Introduction

COMPASS 1 is a two-stage magnetic spectrometer designed to investigate the structure of hadrons using high-intensity muon and hadron beams from the SPS accelerator at CERN. The main component of the small-area tracking are ten GEM stations, each made up out of two detectors, one with horizontal and vertical readoutstrips, the second one rotated by 45° with respect to the first. For the operation in high-intensity beams, the central region of the GEM detectors can be deactivated to reduce occupancy. The GEM stations are mounted to the centers of the COMPASS large area trackers (straws or MWPCs) to provide high spatial resolution close to the beam. The Gas Electron Multiplier2 is a thin (50 /an) Cu-clad kapton foil perforated with holes of 70 /xm diameter at a pitch of 140 /Ltm. Application of a voltage between the two metalized faces of the foil leads to high electric fields in the holes and allows electron amplification. To reach high gains, several such foils can be cascaded.

259

260 2

T h e G E M Detectors

The COMPASS GEM detectors consist of three GEM stages. Since the detectors will be operated in high-intensity hadron beams with a background of heavily-ionizing nuclear fragments, a minimization of the discharge probability is crucial. To this end, the voltage sharing between the three GEM foils is asymmetric, with the highest voltage difference across the topmost foil3. In addition, the foils are subdivided into twelve parallel segments and a circular central sector with 50 mm diameter, which are individually connected to the HV distribution chain. This limits the available energy in case of a discharge. The disc-shaped central sector can be deactivated remotely to permit operation in high-intensity beams. The voltages are applied via a resistor network, so that only one external HV connection is necessary per detector. The two-dimensional orthogonal readout is realized with 768 parallel strips per coordinate at a pitch of 400 /am, the width of the strips having been adjusted to achieve equal charge sharing. The upper strips are 80 pm in width, the lower ones 340 (j,m, down from 350 fan in the first batch of the production to avoid short-circuits. As front-end readout electronics, the APV25-S0 chip 4 developed for the CMS silicon tracker is used. A protection circuit consisting of antiparallel diodes and blocking capacitors shields the chip from large electrical pulses caused by eventual discharges. •.WSLLZ,-,:,

drift honeycomb

jj*

™'*jj

mmEiar

3.0

GEM1 r

2.0

15.8

2.0

JOUt
F.

(1)

where Qpre, Qposti, QPost2- charges collected from first, second and third chamber respectively. Efficiencies for lepton identification was obtained from experiments with beams of electrons (850 MeV) and protons (2.1 GeV). The results are presented in the picture (Figure 3). In addition, the simulation results are enclosed. For the F=2 efficiency for electron identification is about 90% and only about 10% hadrons we misidentified as electrons. In addition a good agreement between simulation and experimental results was achieved.

275

1 electrons 850 MeV

0.5 a

fakes - protons 2.1 GeV

0.1

*

0.05 simulation data experimental data

0.02 0.01

2.2

2.4 2.6 parameter F

2.8

Figure 3. The measured and simulated electron identification and misidentification (fakes) efficiencies as a function of F parameter.

4

Conclusion

The PRE-SHOWER detector provides an effective way of leptons from hadrons separation. The detector together with dedicated stable electronic system 5 allows to perform online lepton recognition in the second level trigger of HADES. References 1. HADES, the New Electron-Pair Spectrometer at GSI J. Friese, for the

HADES collaboration Nucl. Phys. A 654, (1999) 1017c 2. P. Salabura at al., Acta Phys. Pol. B 27(1,2)(1996)421 3. C. Agodi et al., "The Time of Flight Wall for the HADES Spectrometer", IEEE Transaction on Nuclear Science, vol. 45, no. 3, June 1998 4. J. A. Kadyk, Nucl. Instr. And Meth. A 300 (1991) 436-479 5. Development of fast pad readout system for the HADES shower detector, A. Balanda at al., Nucl. Instr. And Meth. A 417 (1998) 360-370

T H E TIME P R O J E C T I O N C H A M B E R FOR T H E CERN-LHC HEAVY-ION E X P E R I M E N T ALICE H.R. SCHMIDT Gesellschaft fur Schwerionenforschung, Planckstr. 1, D-64S91 Darmstadt, Germany E-mail: [email protected] The ALICE TPC is a conventional TPC based on experience with previous TPCs used in heavy ion beams. However, the unpreceeded high particle multiplicities at LHC Pb+Pb collision has led in detail to many innovations in its design and construction. 1

Introduction

It seems that roughly every 5 years, forced by the augmented energy of heavy ions beams, a major Time Projection Chamber (TPC), setting new standards, comes into operation. This is documented in Table 1, which contains a comparison of design parameters of the NA49 2 TPC at the SPS (y/s = 17 GeV) , the STAR 3 TPC at RHIC ( ^ i = 200 GeV) and the projected ALICE * TPC at LHC (y/s = 5500 GeV) . From this table it can be seen that the ALICE TPC exceeds its predecessors in basically all aspects. This is forced by the expected high charged particles multiplicities of up to 8000 per unit rapidity, which is a factor of 5-10 higher than at RHIC. These unpreceeded high multiplicities are a major challenge both to the construction and the operation of the ALICE TPC. In this contribution we will show how the design of the ALICE TPC readout chambers, being basically conservative and based on the NA49 and STAR T P C s , are optimized to be able to handle the high particle load. 2

T P C Layout

The overall acceptance of the TPC is 0.9 < r\ < 0.9. To cover this acceptance the TPC is of cylindrical design with an inner radius of about 80 cm, an outer radius of about 250 cm, and an overall length in the beam direction of 500 cm schematic layout of the ALICE TPC is shown in Fig. 1 The TPC field cage provides a highly uniform electrostatic field in a cylindrical high-purity gas volume to transport primary charges over long distances (2.5 m) towards the readout end-plates. The field configurations is defined by a high-voltage (up to 100 kV) electrode located at the axial centre of the

276

277 Table 1. Comparison of the NA49, the Star and the planned ALICE TPC

parameter No. of channels gas

gas gain field cage drift voltage minimal pad size Luminosity [cm _ l i s " 1 ]

NA"4¥ SPS fixed target 1995 182k NeCC-2 (90-10) (vertex) ArCH4C02 (90-5-5) (main) 2 x 10* (vertex) 5 x 10 3 (main) W = 39"0cm L = 390cm H = 180cm (main) V = 27m3 „ 13.4/19.5 kV 200/175 V / c m 3.5 X 16 m m 2 = Kfi 0 m m 2

»1025

—ATICE LHC collider

STAR RHIC collider 2000 140k

570k

ArCH 4 (90-10)

NeCOa (90-10)

3.6 x 10 3 (inner) 1.3 x 10 3 (outer)

2 x 10 4

L = 420 cm R = 210cm V = 50m3 31 kV 150 V / c m 3.5 x 11.5 m m 2 = 39..S m m 2

2 X 10 2 5

L = 500cm R = 250cm V = 88m3 lOOkV 400 V / c m 4 x 7.5 m m 2 = 30 0 m") 2 0.5 - 1 X 10 2 7

SL-i-'fc

Figure 1. The ALICE TPC showing the central electrode, the field cage.and the end plates

cylinder. As drift gas a mixture of N e C 0 2 , as is currently used in the NA49 (9Q%/10%) and CERES (80%/20%) experiments at the SPS, is chosen. The readout chambers are basically conventional multiwire proportional

278

chambers with cathode pad readout as used in many TPCs before. In detail, their construction, however, requires to overcome significant technical challenges as discussed below. The overall area to be instrumented is 32.5 m 2 . The azimuthal segmentation of the readout plane follows that of the subsequent ALICE detectors, leading to 18 trapezoidal sectors, each covering 20 degrees in azimuth. The radial decrease of the track density leads to changing the requirements for the readout chamber design as a function of radius. Consequently, there will be two different types of readout chambers, the inner and outer chambers. Each outer chamber is further subdivided into two sections with different pad sizes, leading to a triple radial segmentation of the readout plane, with 557568 readout pads in total.

3

T P C challenges

The most obvious negative consequence of a high track density is the corresponding high occupancy of the readout channels. In the following we show that a simple increase in the readout granularity would be of limited help if not accompanied by a number of other measures. 3.1

Pad Size

A sufficient number of pads per charge cluster in terms of position resolution is 2-3. Thus an increase in the number of pads is sensible only if the induced charge from the (point-like) avalanche spread over no more the 2-3 pads. This can be achieved by reducing the distance between anode wire and pad plane. However, at a certain point the distance HV-GND gets critical. There are also other reasons why it makes little sense to decrease the pad size beyond a certain limit: the width of the charge cloud after 250 cm of drift is of the order of mm depending on the choice of the drift gas and voltage. I.e., any reduction of the pad beyond a certain size result in an oversampling of the track without any gain of information. 3.2

Time direction

The situation in time direction is similar: one could think of increasing the frequency of the time sampling, however, as diffusion occurs also in longitudinal direction this would result as well in an oversampling of the pulse. The choice of a shorter shaping time is limited by the fact that below 150-200 ns shaping time the signal/noise ratio becomes critical.

279 3.3

Optimization

From of the above considerations one is left with the following measures to optimize for best performance in a high density environment: 1) minimization of the diffusion, i.e. choice of a "cold" gas (NeC02, 90-10) and a high drift field (400V/cm). 2) choice of a minimal pad area (A = 30 mm 2 ) which still gives a reasonable signal; this implies a) the proper choice of the anode-pad distance (2 mm) to have the desired pad response function (PRF) and b) a high gain, because the faint signal from the small pad needs high amplification. This can be done both by gas and electronic amplification, however, to optimize the signal/noise ratio (S/N^20) a high gas gain (2 xlO 4 ) and low electronic gain (8mV/fC) is preferable. This choice should lead to a number of equivalent noise electrons below 1000. 3) For a given pad area the proper choice of the aspect ratio (4 x 7.5 mm 2 ) will further decrease the number pads occupied by a cluster. The principal reason for this is that the tracks are oriented in a preferred direction, i.e., radially. 4

Long Term Stability

In principle, the long term behavior of gaseous detectors is not testable, as the only halfways realistic test would require an exposure of the chambers at rates and durations comparable to the experimental conditions. For an expected running time of several years this is not possible for obvious reasons. One resorts therefore to short time tests with high intensity exposure to accumulate at least as much charge per unit length anode wire (where the amplification takes place) as in the experiment. In our case we exposed an anode area of about 1 cm 2 with a strong 55 Fe source for about 2000 hrs. The resulting anode current of 25 nA was monitored and found to be stable for the whole measurement period. The corresponding charge/unit length of the anode wire is calculated to be 60 mC. This has to be compared with an estimated accumulated charge of 1.1 mC per cm wire and ALICE year (1 ALICE year = 10 6 s). 5

Space Charge

There are two distinct sources of space charge in the TPC drift volume:

280

a) positive ions from primary ionization by a charged particle, and b) positive ion leaking back from the amplification zone into the drift space. Owing to the much smaller mobility of the ions as compared to the electrons a quasi-stationary positive charge will distort the drift field significantly. While a) is unavoidable and leads to distortions of the tracks of up to 0.5 mm for NeCC>2 as drift gas, b) could cause much larger distortions if the ion feedback is not sufficiently blocked by the gate. This is particularly dangerous at the high amplification of 2 x 104 in the present ROC's. First tests on prototype chambers showed indeed that ions leak back into the drift space even with gate closed. Two-dimensional calculations of the field configuration revealed the ion-leaks were located at the radial borders of the chamber, i.e., at the discontinuities of the otherwise regular gating grid structure. To circumvent the problem electrostatic "shims" were introduced to optimize the field geometry. The gating inefficiency was assessed by measuring the drift electrode current as a function of the gating offset voltage. At high gating offset voltage the measurement was limited by the sensitivity of the ammeter of « 10 pA. An upper limit of 0.5 x 1 0 - 4 for the gating inefficiency was deduced. This, together with an amplification of 2 x 104 results in less than 20 ions/cm track coming from the amplification and is of the same order of magnitude as the ions from primary ionization. 6

High Rate

So far, previous TPC's have not yet been operated both at high gain and at high track density. It is thus questionable whether under those conditions the chamber can be operated stably at all. A first test was performed at GSI employing a TPC readout chamber formerly used in the NA35 experiment. The chamber was irradiated with secondaries from a 12 C beam hitting a thick target. By varying the target thickness and/or the beam intensity track densities from overlapping events similar LHC P b + P b collisions could be reached. It turned out that the chamber could sustain several tens of fiA anode current without signs of instability. 7

Simulation Results

Even after the optimization steps described above one is left with an occupancy exceeding 50% at the innermost radius for an assumed multiplicity of dN/dy = 8300 plus background ( « 30%). Previous experience from the NA49 experiment demonstrated that the tracking efficiency is reduced dramatically

281

for occupancies above 20%. The situation, however, is different for a fixed target experiment as NA49 and an experiment in collider geometry where the track density decreases quadratically with the radius. The ALICE tracking group has adopted novel tracking algorithms which are based on local methods, i.e., the tracking starts at the outer parts of the TPC and proceeds to smaller radii. No global track model is needed in this case. Employing Kalman filtering leads to an acceptable efficiency, i.e., of 88% of all recognizable track are found with only 2% fake tracks. At present, the momentum resolution is evaluated for the TPC only, i.e., the connection to the other tracking detectors - ITS at small radii and TRD at large radii - is not included in the tracking algorithms. The momentum resolution Ap/p at lGev/c is found to be 2.4%, which is close to the expectation of the Technical Design Report 1. For high momentum tracks with p > 5GeV/c the resolution is at present > 14%, clearly not good enough for high pt physics. However, with the additional information from the other tracking detectors is it expected that the resolution for high momentum track is well below 5% 4 . The simulations yield a dE/dx resolution of 8-9% in the high track environment, while the resolution of a single, isolated track is « 5%, which is close to the optimum. Thus the particle identification properties of the TPC are as good as they can possibly be under the given circumstances. 8

Summary

We have shown that the measures taken to optimize the TPC readout chambers will allow to operate the ALICE TPC even under the highest anticipated particle load. The simulated performance indicates that momentum and dE/dx will be sufficient the reach the physics goals as formulated in 1. References 1. ALICE Collaboration, Technical Design Report, CERN/LHCC 2000-001 2. S. Afanasev et al., The NA49 Large Acceptance Hadron Detector, Nucl.Instrum.Meth. A430:210-244, 1999 3. H. Wieman, Recent Developments in TPC Technology and the STAR TPC at RHIC, this conference 4. ALICE Collaboration, Technical Design Report, CERN/LHCC 2001-021, p. 156

CATHODE STRIP CHAMBERS DATA ANALYSIS I. GOLUTVIN, Y. KIRYSHIN, K. MOISSENZ, P. MOISSENZ, S. MOVCHAN, A. ZARUBIN Joint Institute for Nuclear Research, 6 Joliot Curie, Dubna, Moscow reg., 141980, E-mail: [email protected]

Russia

Main cathode strip chambers (CSC) data analysis tasks are: calibration, reconstruction of the transmission function (transformation of CSC readout information to the coordinate of particle), track finding, optimal track parameters estimation and alignment. Methods for solution of these tasks (excepting calibration) are described. The influence of CSC geometrical parameters, sampling time and a number of signal readout samples, uncorrelated background, overflows and magnetic field to the spatial resolution are analysed. Proposed methods and algorithms are useful for the data analysis as well as for detector optimisation.

1

Introduction

CSC has been chosen for the endcap muon system of the Compact Muon Solenoid (CMS) because it is capable to provide precise spatial and timing resolution in presence of a high magnetic field and high particle background rate.1,2 Each CSC module consists of six layers to provide robust pattern recognition for background rejection and efficient matching of external muon tracks to internal track segments. The layer is a multiwire proportional chamber in which one cathode plane is segmented into strips running across wires. Radial strips and wire groups provide a natural

2) with collected charge greater then noise and bounded (left and right) by strips without charge. Let Qc is a charge of the central strip in the cluster, Q is a charge of the left neighbour strip, Q,. is a charge of the right neighbour strip, x is the distance between muon and centre of the strip with Qc. There are two methods for x calculation: ratio method and fitting method.5

282

283

2.1

The Ratio Method

Let

a=

'

'

where W is a strip width, then x(ax) =

-w

+ WJp(a)d(a)

0)

(for the uniform distribution of x across the strip!) where p(a) is probability density distribution from a. Spatial resolution is

_ ORATIO-

gV(&-a) 2+ (a-Q) 2 +(a-&) r &'(&-&)-&'(&-a)-a'(&-e,)

(2)

where o is standard deviation of readout channel noise. For x = ±.5-W or x = 0

ORATIO=

vrf^V

(3)

where Q is the cluster charge, Q-q = Q r (q is a fraction of strip charge (from Q) vs distance between strip centre and muon). Sometimes Q,. is not correctly measured, due to the dynamic range of electronics. In this case the following formula can be used6 Qr-Qi

a=

^—=

(A\

and then spatial resolution is

V2(rV(8f + Q, -8 r + , -6,_,)2 +(6f - g ; ) 2 ORATIO

2.2

(Q;-Q'r)(QiM+QlJ

+

(Qr-Ql)(Q'r+l+Q;_1) + 2a'rQl-2Q'lQ,

( )

The Fitting Method

According to the fitting method muon coordinate x and cluster charge Q are calculated from the minimum of F{x,Q) = Jd{^-Qq(x)f

(6)

where n is a number of strips in the cluster, i is strip number in cluster, Qjexp is measured charge in strip i, qi is theoretical value of the measured charge in strip i . Spatial resolution is

284

(7)

One can see that both methods give the similar spatial resolution in the centre of strip and between strips for the narrow clusters. Radial structure of the strips gives the possibility to calculate x, Q, R, and A from the minimum of F(x,e,tf,A) = j r f e M P - 2 [ ( l - 2 ( * , + *2))