Design, performance, and calibration of CMS hadron-barrel calorimeter wedges

Available on CMS information server CMS NOTE-2008/020

The Compact Muon Solenoid Experiment

Mailing address: CMS CERN, CH-1211 GENEVA 23, Switzerland

CMS Note

24 June 2008

Design, Performance, and Calibration of the CMSHadron-Outer Calorimeter

S. Abdullin, V. Abramov, B. Acharya, M. Adams, N. Akchurin, U. Akgun, E. Albayrak, E. W. Anderson, G.Antchev, M. Arcidy, S. Ayan, S. Aydin, M. Baarmand, K. Babich, D. Baden, M. N. Bakirci, Sudeshna Banerjee,Sunanda Banerjee, R. Bard, V. Barnes, H. Bawa, G. Baiatian, G. Bencze, S. Beri, L. Berntzon, V. Bhatnagar, A.Bhatti, A. Bodek, S. Bose, T. Bose, H. Budd, K. Burchesky, T. Camporesi, K. Cankocak, K. Carrell, S. Cerci, S.

Chendvankar, Y. Chung, W. Clarida, L. Cremaldi, P. Cushman, J. Damgov, P. de Barbaro, P. Debbins, M.Deliomeroglu, A. Demianov, T. de Visser, P. V. Deshpande, J. Diaz, L. Dimitrov, S. Dugad, I. Dumanoglu , F.

Duru, I. Efthymiopoulos, J. Elias, D. Elvira, I. Emeliantchik, S. Eno, A. Ershov, S. Erturk, E. Eskut, A. Fenyvesi,W. Fisher, J. Freeman, V. Gaultney, H. Gamsizkan, V. Gavrilov, V. Genchev, Y. Gershtein, S. Gleyzer, I. Golutvin,P. Goncharov, T. Grassi, D. Green, A. Gribushin, B. Grinev, A. Gurtu, A. Murat Guler, E. Gulmez, K. Gumus, T.

Haelen, S. Hagopian, V. Hagopian, M. Hashemi, J. Hauptman, E. Hazen, A. Heering, N. Ilyina, D. Ingram, E.Isiksal, C. Jarvis, C. Jeong, K. Johnson, V. Kaftanov, V. Kalagin, A. Kalinin, D. Karmgard, S. Kalmani, M. Kaur,M. Kaya, O. Kaya, A. Kayis-Topaksu, R. Kellogg, A. Khmelnikov, H. Kim, I. Kisselevich, O. Kodolova, J. Kohli,V. Kolossov, A. Korablev, Y. Korneev, I. Kosarev, S. Koylu, L. Kramer, A. Krinitsyn, A. Krokhotin, V. Kryshkin,

S. Kuleshov, A. Kumar, S. Kunori, P. Kurt, A. Laasanen, V. Ladygin, G. Landsberg, A. Laszlo, C. Lawlor, D.Lazic, S.-W. Lee, L. Levchuk, S. Linn, D. Litvintsev, L. Litov, L. Lobolo, S. Los, V. Lubinsky, V. Lukanin, Y. Ma,

E. Machado, J. Mans, M. Maity, G. Majumder, P. Markowitz, V. Mossolov, G. Martinez, K. Mazumdar, J. P.Merlo, H. Mermerkaya, G. Mescheryakov, A. Mestvirishvili, M. Miller, A. Moeller, M. Mohammadi-Najafabadi,

P. Moissenz, N. Mondal, P. Nagaraj, E. Norbeck, J. Olson, Y. Onel, G. Onengut, C. Ozkan, H. Ozkurt, S.Ozkorucuklu, S. Paktinat, A. Pal, M. Patil, A. Penzo, S. Petrushanko, A. Petrosyan, V. Pikalov, S. Piperov, V.Podrasky, A. Polatoz, A. Pompos, S. Popescu, C. Posch, A. Pozdnyakov, W. Qian, R. M. Ralich, L. Reddy, J.

Reidy, R. Ruchti, E. Rogalev, Y. Roh, J. Rohlf, A. Ronzhin, A. Ryazanov, G. Safronov, D. A. Sanders, C. Sanzeni,L. Sarycheva, B. Satyanarayana, I. Schmidt, S. Sekmen, S. Semenov, V. Senchishin, S. Sergeyev, M. Serin, R.Sever, J. B. Singh, A. Sirunyan, A. Skuja, S. Sharma, B. Sherwood, N. Shumeiko, V. Smirnov, K. Sogut, N.

Sonmez, P. Sorokin, M. Spezziga, R. Stefanovich, V. Stolin, K. Sudhakar, L. Sulak, I. Suzuki, V. Talov, K. Teplov,R. Thomas, H. Topakli, C. Tully, L. Turchanovich, A. Ulyanov, A. Vanini, I. Vankov, I. Vardanyan, F. Varela, M.

Vergili, P. Verma, G. Vesztergombi, R. Vidal, A. Vishnevskiy, E. Vlassov, I. Vodopiyanov, I. Volobouev, A.Volkov, A. Volodko, L. Wang, M. Wetstein, D. Winn, R. Wigmans, J. Whitmore, S. X. Wu, E. Yazgan, T. Yetkin,

P. Zalan, A. Zarubin, M. Zeyrek

Abstract

The CMS hadron calorimeter is a sampling calorimeter with brass absorber and plastic scintillatortiles with wavelength shifting fibres for carrying the light to the readout device. The barrel hadroncalorimeter is complemented with an outer calorimeter to ensure high energy shower containment inthe calorimeter. Fabrication, testing and calibration of the outer hadron calorimeter are carried out

keeping in mind its importance in the energy measurement of jets in view of linearity and resolution.It will provide a net improvement in missing measurements at LHC energies. The outer hadroncalorimeter will also be used for the muon trigger in coincidence with other muon chambers in CMS.

keeping in mind its importance in the energy measurement of jets in view of linearity and resolution.It will provide a net improvement in missing measurements at LHC energies. The outer hadroncalorimeter will also be used for the muon trigger in coincidence with other muon chambers in CMS.

EPJ manuscript No.(will be inserted by the editor)

Design, Performance, and Calibration of the CMS Hadron-OuterCalorimeter

CMS HCAL Collaborations

S. Abdullin12, V. Abramov14, B. Acharya8, N. Adam35, M. Adams24, N. Akchurin28, U. Akgun27, E. Albayrak27,E. W. Anderson21, G. Antchev23, M. Arcidy23, S. Ayan56,73, S. Aydin16, T. Aziz8, M. Baarmand29, K. Babich11,D. Baden25, M. N. Bakirci16, Sudeshna Banerjee8, Sunanda Banerjee8, R. Bard24, V. Barnes39, H. Bawa7,G. Baiatian1, G. Bencze5, S. Beri7, L. Berntzon28, V. Bhandari7, V. Bhatnagar7, A. Bhatti32, A. Bodek37,S. Bose8, T. Bose36, H. Budd37, K. Burchesky24, T. Camporesi15, K. Cankocak18,27, K. Carrell28, S. Cerci16,S. Chendvankar8, Y. Chung37, W. Clarida27, L. Cremaldi34, P. Cushman31, J. Damgov3,22, P. de Barbaro37,P. Debbins27, M. Deliomeroglu18, A. Demianov13, T. de Visser15, P. V. Deshpande8, J. Diaz30, L. Dimitrov3,S. Dugad8, I. Dumanoglu16, F. Duru27, I. Efthymiopoulos15, J. Elias22, D. Elvira22, I. Emeliantchik2, S. Eno25,A. Ershov13, S. Erturk16,41, S. Esen36, E. Eskut16, A. Fenyvesi6, W. Fisher35, J. Freeman22, S. N. Ganguli8,V. Gaultney30, H. Gamsizkan17, V. Gavrilov12, V. Genchev3, S. Gleyzer38, I. Golutvin11, P. Goncharov14, T. Grassi25,D. Green22, A. Gribushin13, B. Grinev20, M. Guchait8, A. Gurtu8, A. Murat Guler17, E. Gulmez18, K. Gumus28,T. Haelen37, S. Hagopian38, V. Hagopian38, V. Halyo35, M. Hashemi9, J. Hauptman21, E. Hazen23, A. Heering31,23,A. Heister23, A. Hunt35, N. Ilyina12, D. Ingram27, E. Isiksal18,42, C. Jarvis25, C. Jeong28, K. Johnson38, J. Jones35,V. Kaftanov12a, V. Kalagin11, A. Kalinin14, S. Kalmani8, D. Karmgard33, M. Kaur7, M. Kaya18,43, O. Kaya18,43,A. Kayis-Topaksu16, R. Kellogg25, A. Khmelnikov14, H. Kim28, I. Kisselevich12, O. Kodolova13, J. Kohli7,V. Kolossov12, A. Korablev14, Y. Korneev14, I. Kosarev11, L. Kramer30, A. Krinitsyn14, M. R. Krishnaswamy8,A. Krokhotin12, V. Kryshkin14, S. Kuleshov12, A. Kumar7, S. Kunori25, A. Laasanen39, V. Ladygin11, E. Laird35,G. Landsberg36, A. Laszlo5, C. Lawlor23, D. Lazic23, S. W. Lee28, L. Levchuk19, S. Linn30, D. Litvintsev12,22,L. Lobolo30, S. Los22, V. Lubinsky20, V. Lukanin14, Y. Ma31, E. Machado23, M. Maity8, G. Majumder8, J. Mans35,31,D. Marlow35, P. Markowitz30, G. Martinez30, K. Mazumdar8, J. P. Merlo27, H. Mermerkaya29, G. Mescheryakov11,A. Mestvirishvili27, M. Miller27, A. Moeller27, M. Mohammadi-Najafabadi9, P. Moissenz11, N. Mondal8, V. Mossolov2,P. Nagaraj8, V. S. Narasimham8, E. Norbeck27, J. Olson27, Y. Onel27, G. Onengut16, C. Ozkan17, H. Ozkurt16,S. Ozkorucuklu18,44, F. Ozok27, S. Paktinat9, A. Pal5, M. Patil8, A. Penzo10, S. Petrushanko13, A. Petrosyan11,V. Pikalov14, S. Piperov3,22, V. Podrasky26, A. Polatoz16, A. Pompos39, S. Popescu28, C. Posch23, A. Pozdnyakov12,W. Qian24, R. M. Ralich29, L. Reddy8, J. Reidy34, E. Rogalev11, Y. Roh28, J. Rohlf23, A. Ronzhin22, R. Ruchti33,A. Ryazanov14, G. Safronov12, D. A. Sanders34, C. Sanzeni26, L. Sarycheva13, B. Satyanarayana8, I. Schmidt27,S. Sekmen17, S. Semenov12, V. Senchishin20, S. Sergeyev22, M. Serin17, R. Sever17, B. Singh7, J. B. Singh7,A. Sirunyan1, A. Skuja25, S. Sharma8, B. Sherwood31, N. Shumeiko2, V. Smirnov11, K. Sogut16,45, N. Sonmez18,46,P. Sorokin19, M. Spezziga28, R. Stefanovich2, V. Stolin12, K. Sudhakar8, L. Sulak23, I. Suzuki22, V. Talov14,K. Teplov13, R. Thomas28, S. Tonwar8, H. Topakli16, C. Tully35, L. Turchanovich14, A. Ulyanov12, A. Vanini36,I. Vankov3, I. Vardanyan13, F. Varela23, M. Vergili16, P. Verma8, G. Vesztergombi5, R. Vidal22, A. Vishnevskiy11,E. Vlassov15,12, I. Vodopiyanov29, I. Volobouev28, A. Volkov14, A. Volodko11, L. Wang25, J. Werner35, M. Wetstein25,D. Winn26, R. Wigmans28, J. Whitmore22, S. X. Wu23, E. Yazgan28, T. Yetkin27, P. Zalan5, A. Zarubin11, andM. Zeyrek17

1 Yerevan Physics Institute, Yerevan, Armenia2 NCPHEP, Minsk, Belarus3 Institute for Nuclear Research and Nuclear Energy, Bulgarian Academy of Science, Sofia, Bulgaria4 Sofia University, Sofia, Bulgaria5 KFKI-RMKI, Research Institute for Particle and Nuclear Physics, Budapest, Hungary6 ATOMKI, Debrecen, Hungary7 Panjab University, Chandigarh, 160 014, India8 Tata Institute of Fundamental Research, Mumbai, India9 Institute for Studies in Theoretical Physics ... Sharif University of Technology, Tehran, Iran

10 Universita di Trieste e Sezione dell’ INFN, Trieste, Italy11 JINR, Dubna, Russia

a deceased

2

12 ITEP, Moscow, Russia13 Moscow State University, Moscow, Russia14 IHEP, Protvino, Russia15 CERN, Geneva, Switzerland16 Cukurova University, Adana, Turkey17 Middle East Technical University, Ankara, Turkey18 Bogazici University, Istanbul, Turkey19 KIPT, Kharkov, Ukraine20 Single Crystal Institute, Kharkov, Ukraine21 Iowa State University, Ames, IA, USA22 Fermi National Accelerator Laboratory, Batavia, IL, USA23 Boston University, Boston, MA, USA24 University of Illinois at Chicago, Chicago, IL, USA25 University of Maryland, College Park, MD, USA26 Fairfield University, Fairfield, CT, USA27 University of Iowa, Iowa City, IA, USA28 Texas Tech University, Lubbock, TX, USA29 Florida Institute of Technology, Melbourne, FL, USA30 Florida International University, Miami, FL, USA31 University of Minnesota, Minneapolis, MN, USA32 Rockefeller University, New York, NY, USA33 University of Notre Dame, Notre Dame, IN, USA34 University of Mississippi, Oxford, MS, USA35 Princeton University, Princeton, NJ, USA36 Brown University, Providence, RI, USA37 University of Rochester, Rochester, NY, USA38 Florida State University, Tallahassee, FL, USA39 Purdue University, West Lafayette, IN, USA40 Now at: University of Pennsylvania, Philladelphia, PA,USA41 At: Nigde University, Nigde, Turkey42 At: Marmara University, Istanbul, Turkey43 At: Kafkas University, Kars, Turkey44 At: Suleyman Demirel University, Isparta, Turkey45 At: Mersin University, Mersin, Turkey46 At: Izmir Yuksek Teknoloji Enstitusu, Izmir, Turkey

Received: date / Revised version date

Abstract. The Outer Hadron Calorimeter (HCAL HO) of the CMS detector is designed to measure theenergy that is not contained by the barrel (HCAL HB) and electromagnetic (ECAL EB) calorimeters.Due to space limitation the barrel calorimeters do not contain completely the hadronic shower and anouter calorimeter (HO) was designed, constructed and inserted in the muon system of CMS to measurethe energy leakage. Testing and calibration of the HO was carried out in a 300 GeV/c test beam thatimproved the linearity and resolution. HO will provide a net improvement in missing ET measurements atLHC energies. Information from HO will also be used for the muon trigger in CMS.

3

1 Introduction

The Compact Muon Solenoid (CMS) experiment [1] at theLarge Hadron Collider (LHC) at CERN is a multi-purposedetector optimised to look for signatures of the Higgs bo-son, super-symmetric particles and other new physics pro-cesses in pp collisions at a centre of mass energy of 14TeV. Signature for production of many of these new parti-cles is large missing transverse energy, ET. The momentaof all charged particles are measured using a high reso-lution tracker in a 4 T solenoidal magnetic field, whileenergies of all particles, charged as well as photons andneutral hadrons, are measured using electromagnetic andhadron calorimeters. Key elements in improving the miss-ing transverse energy measurement are hermiticity of thecalorimeter, excellent energy resolution and small energyleakage.

Fig. 1. Location of the hadron calorimeter in and aroundthe CMS magnet. HCAL HB and HCAL HE are the hadroncalorimeters. ECAL EB and ECAL EE are the electromagneticcalorimeters and PS is pre-shower detector.

Figure 1 shows the longitudinal quarter view of theCMS detector. The hadron calorimeter [2], located behindthe tracker and the electromagnetic calorimeter as seenfrom the interaction point, is a sampling calorimeter madeof copper alloy 1 absorber and plastic scintillators. Thebarrel hadron calorimeter (HB) has a polygonal structure.It is made out of two halves each 4.33 m long. A half barrelis made of 18 wedges each subtending 20◦ in azimuth φ.The HB has 16 scintillator layers designed to be hermeticwith minimal uninstruemented gaps. HB extends up to |η|≈ 1.4 (η = − ln tan θ

2, where θ is the polar angle).

HB is radially restricted between the outer extent ofthe electromagnetic calorimeter (R = 1.77 m) and themagnet cryostat and coil (R = 2.95 m). This constrainsthe total amount of material for the calorimeters EB plusHB to about 7.8 interaction lengths λI at η = 0 as can beseen in Figure 2.

1 Cartridge brass # 260 with 70% Copper and 30% Zinc

η0 0.5 1 1.5 2 2.5

)λM

ater

ial B

udge

t (

02468

101214161820

With HO

Without HO

Before HCAL

Fig. 2. Number of interaction lengths of the CMS calorimeteras a function of η. The two shaded regions correspond to thesetups with or without the outer hadron calorimeter (HO).

Extensive Monte Carlo simulations and test beam stud-ies were carried out [3] to define the parameters of thehadron calorimeter. Figure 3 shows the fraction of eventsthat are missing more than one third of the energy for atest beam of 300 GeV/c for HB alone. Adding the trackerand EB, at 300 GeV/c nearly 3% of pion events gives riseto missing energy corresponding to 100 GeV. Small num-ber of events with large leakage of energy lead to largefluctuation on an event by event basis and cannot be cor-rected offline. This unmeasured energy is the primary jus-tification for the outer hadron (HO) calorimeter.

0

1

2

3

4

5

6

7

8

5 6 7 8 9 10 11 12

Fig. 3. Fraction of 300 GeV/c pion data with reconstructedenergy less than 200 GeV (approximately 3σ below the mean,and corresponding to more than 100 GeV missing energy) ver-sus total absorber depth obtained from 1996 test beam dataat 3 T magnetic field. The total λint includes the contributionof the electromagnetic calorimeter.

4

2 The Outer Hadron Calorimeter (HO):Constraints and Expectation

The EB plus HB inside the 4 T cryostat and solenoidis relatively thin. To ensure adequate sampling depth forη < 1.4, the hadron calorimeter is extended outside thecryostat. The HO will utilize the cryostat and solenoidcoil as an additional absorber (1.4 λint/sinθ) and will alsoidentify and measure the late starting showers.

Outside the cryostat the magnetic field is returnedthrough an iron yoke designed in the form of five 2.536 mwide (along z-axis) rings. The HO is placed as the first sen-sitive layer in each of these five rings. The rings are iden-tified by numbers −2, −1, 0, +1 and +2. The numberingincreases with z and the nominal central z positions of thefive rings are respectively −5.342m, −2.686m, 0, +2.686mand +5.342m. Since the EB plus HB at the central ring(ring 0) has the smallest value of λI, two layers of HOscintillators were placed on either side of a 19.5 cm thickpiece of iron (the tail catcher iron) at radial distances of3820 mm and 4070 mm, respectively. All other rings havea single HO layer at a radial distance of 4070 mm. Thetotal depth of the calorimeter system is thus extended toa minimum of 10λint (see Figure 2) except at the barrel-endcap boundary region.

The HO is constrained by the geometry and construc-tion of the muon system. Figure 1 shows the position of theHO layers in the rings of the muon stations. The segmen-tation of these detectors closely follows that of the bar-rel muon system. Each ring has the 12 identical φ-sectors.The 12 sectors are separated by 75 mm thick stainless steelbeams which hold successive layers of iron of the returnyoke. The space between successive muon rings in the ηdirection and also the space occupied by the stainless steelbeams in the φ direction cause some reduction in the ac-tive area of the HO. In addition, the space occupied by theCMS ‘chimneys’ in the vertical sector of ring −1 and +1are also not available for the HO as well. The chimneysare used for the cryogenic transfer lines and power cablesof the magnet system. Finally, the mechanical structuresneeded to position the scintillator trays further constrainsHO along φ.

In the radial direction, space for HO is limited to only16 mm in the iron yoke, as there are aluminium honeycomb support structures for the muon detector system. Inaddition, the HO modules are independently supportedfrom the steel beams located on either side of each φ sec-tor. The thickness and position of the iron ribs in the yokestructure further constrains the shape and segmentationof the HO.

The sizes and positions of the HO 1.0 cm thick scin-tillator tiles (BC408) roughly map onto the HB towers ofgranularity 0.087× 0.087 in η and φ. The HO consists ofone (rings ±1 and ±2) or two (ring 0) layers of scintillatortiles located in front of the first layer of the barrel muondetector. Scintillation light from the tiles is collected us-ing multi-clad Y11 Kuraray wave-length shifting (WLS)fibres, of diameter 0.94 mm, and transported to the photodetectors located on the structure of the return yoke by

splicing a multi-clad Kuraray clear fibre (also of 0.94 mmdiameter) with the WLS fibre. In order to simplify instal-lation of the HO, the scintillator tiles are packed into asingle unit called a tray. Each tray corresponds to one φslice (∼5◦ wide in φ). However, along the z (η) direction, atray covers the entire span of a muon ring. Figure 4 showsa schematic view of a HO tray where one tile is mappedto a tower of the HB and the optical cable from the trayis connected to the readout box.

Fig. 4. Schematic view of a HO tray shown with individualtiles and the corresponding grooves for WLS fibres. Each tile ismapped to a tower of HB. Optical fibres from the tray extend tothe decoder box which contains the photo-detector and readoutelectronics.

The physics impact of HO has been studied [4] us-ing the simulation tool CMSIM developed for the CMSdetector. The simulation was carried out for single pionsat specific η values. The deposited energy in EB plus HBshow a deficit in energy measurement starting at 70 GeV/cat η = 0 (ring 0). The results with HO are more Gaus-sian in energy distribution indicating that the addition ofthe HO properly recovers the energy leakage. Even withHO in ring 0 there is some small amount of energy leak-

5

age. There is evidence of leakage beyond HO in ring 1 butwith reduced intensity. The amount of leakage in ring 2 isfound to be negligible.

A critical signature of several new physics signals is theevidence of unobserved particle(s), whether it is supersym-metry, or leptoquark or just plain production of dark mat-ter. CMS detects unobservable particle(s) by measuringmissing transverse energy ET and will use it as part of thetrigger and data analysis. Any energy that escapes obser-vation will increase the background making data analysisharder. HO has been designed to reduce the unobservedenergy background. QCD events also have missing ET dueto the production of neutrinos. The cross section of QCDevents, where at least one particle has ET above 500 GeV,is estimated to be several pb. Clearly these events will beaffected due to leakage of energy in the hadron calorime-ter, and the HO would help to decrease the backgroundand improve the energy measurement. Figure 5 shows theaccepted integrated cross section due to QCD processesfor missing ET above a certain value measured with andwithout HO. It is clear that the inclusion of HO reducesthe background cross section by a factor of 1.5 or more formoderate ET values.

10-2

10-1

1

0 200 400 600 800

with HOwithout HO

Missing ET Cut (GeV)

σ (p

b)

Fig. 5. Accepted integrated cross section above threshold as afunction of missing ET when the missing ET is measured withand without the HO.

3 HO Module: Design and Assembly

3.1 Mechanical Design

HO is physically divided into 5 rings in η conforming tothe muon ring structure. The rings are numbered −2, −1,

0, +1 and +2 with increasing η. Each ring of the HO is di-vided into 12 identical φ sectors (numbered 1 to 12 count-ing clockwise starting from 9 o’clock position) and eachsector has 6 slices (numbered 1 to 6 counting clockwise)in φ. The φ slices of a layer are identical in all sectors. Thewidths of the slices along φ are given in Table 1. In each φslice, there is a further division along η. The smallest scin-tillator unit is called a tile. The scintillator tiles in eachφ sector belong to a plane. Perpendicular distance of thisplane from the z-axis is 3.82 m for layer 0 and 4.07 m forlayer 1. The tiles in each φ slice of a ring are mechanicallyheld together in the form of a tray (details in Section 3.2).

Table 1. Dimension of tiles along φ for different trays. Eachtray corresponds to one φ-slice in a φ sector.

Tray # Ring 0 Ring ±1, ±2Layer 0 Layer 1 Layer 1

Tray 1 272.6 mm 298.6 mm 315.6 mmTray 2 341.6 mm 362.6 mm 364.6 mmTray 3 330.6 mm 350.6 mm 352.6 mmTray 4 325.6 mm 345.6 mm 347.6 mmTray 5 325.6 mm 345.6 mm 347.6 mmTray 6 266.6 mm 290.6 mm 404.6 mm

Both layers of ring 0 have 8 η-divisions (i.e. 8 tiles ina tray): −4, −3, −2, −1, +1, +2, +3, +4. Ring 1 has 6divisions: 5, · · ·, 10 and ring 2 has 5 divisions: 11, · · ·, 15.Ring −1 and ring −2 have the same number of divisions asrings 1 and 2 but with negative indices. The η-dimensionsof any tile with negative tower number is the same as theone with positive number. Tile dimensions along η for thetowers are shown in Table 2.

Table 2. HO tile dimensions along η for different rings and lay-ers. The tile sizes, which are constrained by muon ring bound-aries, are mentioned in parenthesis.

Tower ηmax Length Tower ηmax Length(mm) (mm)

Ring 0 Layer 0 Ring 0 Layer 11 0.087 331.5 1 0.087 351.22 0.174 334.0 2 0.174 353.83 0.262 339.0 3 0.262 359.24 0.326 (248.8) 4 0.307 (189.1)

Ring 1 Layer 1 Ring 2 Layer 15 0.436 391.5 11 0.960 420.16 0.524 394.2 12 1.047 545.17 0.611 411.0 13 1.135 583.38 0.698 430.9 14 1.222 626.09 0.785 454.0 15 1.262 (333.5)

10 0.861 (426.0)

Length of a full tray is 2510 mm whereas the shortertrays, the sizes of which are constrained because of thechimney (trays 4 and 5 in sector 4 of ring +1 and trays3, 4, 5 and 6 in sector 3 of ring −1), are 2119 mm long.The shorter trays are constructed without the tile corre-

6

sponding to tower number ±5. Because of the constraintsimposed by the gap between the two rings, a part of tower# ±4, which falls in ring ±1 (tower ±4 is restricted onlyto ring 0) is merged with tower # ±5.

There is a maximum of 40 mm of clearance for plac-ing the HO trays together with the mechanical structures.Each tray with all its packing is expected to have a thick-ness of 15.44 mm. This necessitated a space of 17.0+0.2

−0.0

mm for inserting the trays. The total weight of a φ sector(six trays) is about 30.0 kg. To support this weight, a pairof aluminium honeycomb panels (10 mm thick on the bot-tom and 6 mm thick at the top) are clamped between threealuminium C-channels on the two sides and one at the endwhile the other end is kept open for inserting the trays.A set of twelve 20 mm × 20 mm aluminium C-channelsare glued on the inner side at a fixed separation for in-serting the trays. The separation is properly adjusted forthe appropriate tray width. All the panels are 2530+5

−0 mmlong. The widths of the three types of panels (layer 0 and1 of ring 0 and layer 1 of rings ±1, ±2) are respectively1920+2

−0 mm, 2050+2−0 mm and 2190+2

−0 mm. Panels for rings±1 and ±2 also carry cooling tubes on one side. Theseserve as thermal screens to prevent heat percolation tothe RPC/muon chamber assembly. The cooling tubes aremade of copper and have inner and outer diameter of 6mm and 8 mm. These are glued to the 6 mm thick honey-comb surface using araldite and fixed with copper collars.For rings ±1, a special panel is built to accommodate theshort trays due to the chimneys.

3.2 Trays

All tiles in each φ slice of a sector are grouped together inthe form of a tray. Figure 6 (a) shows a schematic view ofa HO tray. Each tray contains 5 tiles in rings ±2; 6 tilesin rings ±1 and 8 tiles in ring 0. The edges of the tilesare painted with Bicron reflecting white paint for betterlight collection as well as isolating the individual tiles ofa tray. Further isolation of tiles is achieved by inserting apiece of black tedlar strip in between the adjacent tiles.The tiles in a tray are covered with a single big piece ofwhite, reflective tyvek paper. Then they are covered withblack tedlar paper to prevent light leak. This package isplaced between two black plastic plates for mechanicalstability and ease of handling. The top plastic cover is2 mm thick and the bottom one is 1 mm thick. Figure 6(b) shows a cross section of a tray to illustrate the differentcomponents.

The 2 mm plastic sheet on the top has 1.6 mm deepchannels grooved on it (on the outer side) to route thefibres from individual tiles to an optical connector placedin a groove at the edge of the tray. A 1.5 mm wide straightgroove runs along the edge of the top cover to accommo-date a stainless steel tube. This is used for the passage of aradioactive source for calibrating the modules. Each con-nector has two holes and these are fixed to the scintillator-plastic assembly through matching holes in them. Each φsector in each ring has 6 trays. There are 360 trays forlayer 1 and 72 trays for layer 0.

(a)

(b)

Scintillator (10 mm)

Black Polysterene (2 mm)

Tedler (0.05 mm)Tyvek (0.15 mm)

Tyvek (0.15 mm)Tedler (0.05 mm)Black Polysterene (1 mm)

������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������

���������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������

���������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������

������������������������������������������������������������������������������������������������������������������������������������������

����������������������������������������������������������������������������������������������������������������������������������������

����������������������������������������������������������������������������������������������������������������������������������������

Fig. 6. (a) Layout of scintillator tiles in a typical tray of ring2. (b) Cross section of a HO tray showing the different compo-nents.

3.3 Tiles

Scintillator tiles are made from Bicron BC408 scintillatorplates of thickness 10+0

−1 mm. Figure 7 shows a typical HOscintillator tile. The WLS fibres are held inside the tile ingrooves with a key hole cross section. Each groove has acircular part (of diameter 1.35 mm) inside the scintilla-tor and a neck of 0.86 mm width. The grooves are 2.05mm deep. Each tile has 4 identical σ-shaped grooves, onegroove in each quadrant of the tile. The grooves closelyfollow the quadrant boundary. The corners of the groovesare rounded to prevent damage to the fibre at the bendand to ease fibre insertion. The groove design is slightlydifferent for the tile where the optical connector is placedat the end of the tray. Since the tiles are large, 4 groovesensure good light collection and reduced attenuation oflight.

3.4 Pigtails

The light collected by the WLS fibres inserted in the tilesis transported to photo-detectors. Captive ends of theWLS fibres, which reside inside the groove, are polished,aluminised and protected using a thin polymer coating.The other end of the WLS fibre comes out of the tilethrough a slot made on the 2 mm thick black plastic cover

7

Fig. 7. View of a typical HO tile with WLS fibres inserted inthe 4 σ-grooves.

sheet. To minimize the loss of light in transportation, theWLS fibre (attenuation length of ∼ 1.8 m) is spliced toa clear fibre (attenuation length of ∼ 7.0 m). The clearfibres from each tile follow the guiding grooves on the topplastic to the optical connector at the end. Each tray hastwo optical connectors mounted on one end of the tray.

The pigtails are tested with the help of a specially builtscanner. A UV lamp moves over the pigtail in steps undera computer controlled mechanical stage. At specific inter-vals, the UV lamp shutter opens, the fibres absorb UVphotons and the re-emitted light in each fibre is transmit-ted to the end to a set of photo diodes. Signals from thephoto diodes are then digitized and analyzed to study thecombined transmission loss due to the spliced junctions,connector interface and attenuation. The mean transmis-sion efficiency and its spread σ over the entire productionwas found to be about 95% and 3% respectively.

3.5 Quality Assurance

Each tray is subjected to quality assurance tests using amoving radioactive source and cosmic ray muons. The ra-dioactive source moves in X and Y direction (XY scanner)under computer control and the output is read by a dataacquiring system. There is an additional set of measure-ments using radioactive source on a tip of a moving wirethrough the stainless steel tube fixed to each assembledtray.

XY Scanner.Figure 8 shows the ADC output at several x and y

points over the surfaces of twelve trays belonging to layer 0of ring 0. The collimated source excites all tiles in a similarway. One clearly sees that signal size for tiles close to theconnector is largest and it slowly decreases for tiles withlarger fibre length. The fluctuations among the trays areat a level much below 10% which is used as the acceptancecriterion for the trays.

Wire Source.

Source Position (mm)

Puls

e he

ight

(arb

uni

t)

6.6 6.3 8.6 7.3 8.8 8.2 4.8 7.3σ(%)=

0

500

1000

1500

2000

0 1000 2000 3000

Fig. 8. Scan of a set of 12 fully assembled scintillator trays(of the 12 sectors) belonging to ring 0 layer 0 tray position3 using the XY scanner. The lines on the figures refer to thepeak positions corresponding to the centre of the tiles. Thetiny peak on the right side is due to a calibration tile used ineach set of measurements.

The tiles are also inter-calibrated using radioactivesource technique using a Co

60 source with a strength of5 mCi. Signals are sampled at the rate of 40 MHz anddata are taken in histogramming mode. The mean of sucha histogram represents the response of tile at the givenreel position. While analyzing the wire source data, thereel position is used to find out the correspondence be-tween tile number and readout channel. This method isalso used during the beam test operation.

Cosmic ray muons.The HO trays are expected to supply at least 10 photo

electrons for a minimum ionizing particle (MIP) if a pho-tomultiplier tubes (PMT) were used in the read out. Thecompleted trays have been tested for their responses toMIPs using cosmic rays in a special stand designed forthis purpose. Signal from the tiles are grouped into twobunches (alternate tiles) and the two bunches are read bytwo calibrated green extended PMT’s. The electronic sig-nal is sent to a charge-sensitive ADC. These PMT’s arecalibrated using a precisely controlled LED source to getthe calibration for ADC count per photo electron.

Data are monitored on-line, and there is a calibrationtile equipped with fibre readout which is placed separatelyand is always read out to monitor PMT gain. The pulseheight spectrum is analyzed offline to determine the num-ber of primary photoelectrons produced. Figure 9a showsa typical ADC spectrum. One sees a very clear peak dueto the passage of cosmic muons well separated from thepedestal peak. The pedestal subtracted pulse height (seeFigure 9c) is fit to a Landau distribution and the posi-tion of the mean value of the spectrum together with thecalibration factor provides the number of photoelectronsyielded in a given tile.

Figure 10 shows a plot of average number of photo-electrons obtained using cosmic muons as a function of the

8

(a)

ADC count (0.25pC)

(b)

ADC count (0.25pC)

(c)

ADC count (0.25pC)

0

100

200

0 2500

50

100

0 200

100

200

0 250

Fig. 9. (a) Raw ADC spectrum for tile # 8 for one of thetrays of layer 0, ring 0. (b) A fit to the pedestal part to aGaussian distribution. (c) Pedestal subtracted distribution fitto a Gaussian convoluted by Landau distribution.

0

5

10

15

20

0 1 2 3 4 5 6 7 8 9

8.9 8.1 6.1 4 8.5 7.3 7.1 7.4σ(%)=

Tile Number in a Tray →

# of

photo

elec

tron (

Npe)

→

Fig. 10. Number of photo electrons for the tiles in fully assem-bled scintillator trays of layer 0, ring 0. Tile number is countedfrom the tray end with optical connectors. There are 12 mea-surements for a given tile number coming from 12 trays and themeasurements have a spread between (4.0-8.9)% depending onthe tile position.

tile number in a set of 12 trays belonging to ring 0 layer0 tray. The number of photo electrons is always above 10.The number of photoelectrons are as high as 20 for someof the tiles close to the connector. The RMS spread in thenumber of photo-electrons is less than 10% which is theacceptance criterion for the trays.

Correlation between all the tests (with cosmic muons,source tube and XY scan) are shown in Figure 11. Thesame set of readout system is used for cosmic and wiresource test, whereas a different PMT, optical connectorsare used in the XY-scanner. Statistical error on these mea-sured variables are less than 1%. HO towers will be cal-ibrated in situ using muons from Cosmic Rays or fromcollision events. The strong correlation among the threeset of measurements validates the use of muons in cali-brating the system.

(a)

Cosmic (Npe

)

Sou

rce (volt

)

(b)

Source (volt)

XY (arb

un

it)

(c)

XY (arb unit)

Cosm

ic (N

pe)

1.5

2

10 12 14 16 18 20

450

575

700

1.4 1.6 1.8 2 2.2

10

15

20

450 500 550 600 650 700

Fig. 11. Correlations between number of photo-electron incosmic test and signal peaks in the wire source and XY sourcetest.

4 Test Beam Studies

Prototypes of HO trays were put together with 2 wedgesof HB and a prototype of one sector of endcap hadroncalorimeter module (HE) in the test beam facility at CERN.These prototype trays were exposed to a variety of chargedbeams of hadrons, electrons and muons over a wide energyrange. These tests have been carried out over a number ofyears.

The signals from these detectors are read out usingHybrid Photo Diodes (HPD) which converts the light intoelectrical charges. These are then measured and encodedinto a non-linear digital scale by the charge integrator IC(QIE). The QIE uses LHC clock to divide time into reg-ular bins of 25 ns and measures the accumulated chargein each time bin. Internally, the QIE uses capacitors toaccumulate the charge. The QIE’s use non linear scale sothat the QIE step size is a approximate constant fractionof the QIE value.

9

Arrival time of signals from a scintillator tile to thereadout element has a significant dependence on its loca-tion in η. Energy is measured by summing over a numberof time steps and this η dependent time of arrival makesthe energy measurement rather complex. This effect hasbeen corrected by introducing an appropriate η dependentdelay for each of the towers (see Figure 12).

η0 2 4 6 8 10 12 14 16Av

erag

e Ti

me

(Cha

rge

wei

ghte

d)

12.6

12.8

13

13.2

13.4

13.6

13.8

14 0 < Q < 60 phi=3

0 < Q < 60 phi=4

0 < Q < 60 phi=5

Ring-0

Ring

-2Ring-1

Estimated Timing Response

Poor Signal Tile

Fig. 12. Charge weighted average time slice as a function of η

index of the tiles in HO before the time delay correction. Thesolid line is the estimated time delay as a function of rapidity.The verical axis is in units of 25 nsec.

The HO trays were calibrated using the radioactivewire-source test as well as by exposing them to a beam ofmuons. Figure 13 shows a plot of ADC counts from ring 0,layer 0 tiles as the radioactive source passes along them.Pedestal is determined by fitting a straight line with zeroslope to the part of the signal when radioactive source isfar away from the tile under consideration. It is evidentfrom the figure that there is a leakage of the signal tothe neighbouring tiles when the source is at the edge ofthe tile. So the pedestal subtracted signals from the neigh-bouring tiles are added with proper weight to get the totalsignal. The signal from a tile and that from its two adja-cent ones are then fitted using an iterative procedure toextract the calibration constants. The HPD for HO read-out was operated at voltages 8 kV and 10 kV and separatedatasets were collected for the two settings. Calibrationconstants obtained with these two sets of data show theexpected consistency (see Figure 14) being different by afactor of 1.415.

For the muon runs, the HPD was operating at 10 kV.Muons of energy 150 GeV were pointed at the centre ofeach HO tile and the ADC signal from the tile was usedto determine the muon calibration constant. To minimizenoise, the sum of 4 time slices with the maximum amountof energy deposit is taken as the signal. The sum of fourtime slices away from the signal region is taken as thepedestal. The pedestal and signal peaks are not well sepa-rated in all the tiles (particularly in ring 1). The pedestalvalue is taken as the fitted mean obtained by fitting the

Source Position (mm)700 720 740 760 780 800 820 840

Sign

al (A

rbitr

ary

Scal

e)

0

0.1

0.2

0.3

0.4

Fig. 13. ADC counts from ring 0, layer 0 tiles as the radioac-tive source passes along them.

Tile Index0 5 10 15 20 25

Ratio

(10k

V si

gnal

/8kV

sig

nal)

1.25

1.3

1.35

1.4

1.45

1.5

1.55

2φ

3φ

5φ

6φ

7φ

Fig. 14. Ratio of calibration constant for ring 1 HO tiles inthe test beam set up with the HPD’s being operated at 10 kVand at 8 kV.

pedestal distribution with a Gaussian curve. The peak re-gion of the pedestal subtracted muon signal is then fit-ted with a Gaussian convoluted Landau distribution andthe most probable value of this distribution from the fit istaken as the muon calibration constant for the tile. Figure15 shows correlation plot of calibration constants obtainedthrough the two sets of measurements.

Figure 16 shows energy distributions for a 300 GeVpion beam with only EB + HB in the beam and with EB+ HB + HO in the beam. The measured distributions forhadron beams of energy above 100 GeV are more sym-metric and have smaller widths after adding HO to theenergy calculation. This confirms the expectations fromsimulation studies shown in Section 2.

Weight factor for HO has been found to be 372 MeV/fCfor beam being shot at η = 0.22 (Ring 0) and 407 MeV/fCat η = 0.56 (Ring 1). Energy resolution for π− beam hasbeen measured as a function of beam energy at those twoconfigurations and are shown in Figure 17. The two sets of

10

Source Signal0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8

Muo

n si

gnal

1.5

2

2.5

3

3.5

4

2φRing1 3φRing1 5φRing1 6φRing1 5φRing2 6φRing2

Fig. 15. Calibration constant from muon signal plottedagainst the corresponding value from the wire source analy-sis.

EB+HB+HOEntries 28127

Mean 300.8RMS 30.6

Energy (GeV)0 50 100 150 200 250 300 350 400 450

1

10

210

310

EB+HB+HOEntries 28127

Mean 300.8RMS 30.6

EB+HBEntries 28127

Mean 289.7RMS 38.19

EB+HBEntries 28127

Mean 289.7RMS 38.19

EB+HBEntries 28127

Mean 289.7RMS 38.19

Fig. 16. Energy distribution for a 300 GeV pion beam mea-sured with EB + HB and with EB + HB + HO.

measurements in each plot refer to measurements withoutand with HO. As can be seen in the Figure, the energyresolution improves significantly with the inclusion of HO.Fits to the resolution distribution indicate that the con-stant term improves from 11.2% to 7.8% for Ring 0 andfrom 9.2% to 6.6% for Ring 1 with the inclusion of HO.

The signal size due to penetrating beam is comparedwith the noise level in the HO tiles from the 2002 testbeam studies (see Figure 18). This study has indicatedthat HO will be able to provide signals for minimum ion-izing particles with efficiency better than 90% keeping thenoise level below 20%. CMS uses signal in the ResistivePlate Chambers (RPC) to trigger for muons in the bar-rel as well as in the endcap region. In the barrel (RPCtrigger towers 1-6) and in the overlap region (RPC trigger

Beam Momentum (GeV)50 100 150 200 250 300

RMS/

Mea

n

0.1

0.12

0.14

0.16

0.18

0.2

0.22

0.24

0.26

0.28

Ring 0EB+HB

EB+HB+HO

(a)

Beam Momentum (GeV)50 100 150 200 250 300

RMS/

Mea

n

0.1

0.12

0.14

0.16

0.18

0.2

0.22

0.24

0.26

0.28

Ring 1EB+HB

EB+HB+HO

(b)

Fig. 17. Energy resolution for pions as a function of beamenergy measured with EB + HB and with EB + HB + HO forthe beam being shot at (a) η = 0.22 and (b) η = 0.56

towers 7-9) for the RPC’s, there is a considerable cov-erage by HO. Even with 95% chamber efficiency of theRPC’s, muon trigger efficiency is rather poor for RPCtrigger towers 6-9 (as low as 72%) if only RPC’s are usedin coincidence. This study suggested that HO could be auseful component in muon trigger together with the RPCat these solid angle[7].

As it has been observed for data collected with theradioactive source, size of the signal collected from HOtiles is considerably bigger when the HPD is operated at10 kV as compared to 8 kV. This is also demonstrated in

11

HO ADC counts-10 -5 0 5 10 15 20

Freq

uenc

y

0

500

1000

1500

2000

2500HPD high voltage = 8 kV

Pedestal width = 1.38

Signal peak = 3.19

(a)

HO ADC counts-10 -5 0 5 10 15 20

Freq

uenc

y

0

500

1000

1500

2000

2500

3000 HPD high voltage = 10 KV

Pedestal width = 1.42

Signal peak = 4.19

(b)

Fig. 18. Pedestal peak and muon signal for a ring 2 tile oper-ated with a voltage of (a) 8 kV, (b) 10 kV on the HO HPD.

the HO signals obtained from the 2002 test beam studieswith 225 GeV/c muon beam, as shown in Figure 18. Useof HO will make the trigger efficiency better than 90%over most of the solid angle. Figure 19 shows a plot ofefficiency of muon detection versus purity of the signalobtained from the 2004 test beam studies.

5 Summary

Constrained to lie outside of the inner CMS detector lay-ers (pixels, tracker and electromagnetic calorimeter) andinside the 4 Tesla magnet, the hadron barrel calorimeterwas found to be too thin to effectively absorb high energyhadrons specially near the η = 0 region. This would leadto degraded sensitivity for establishing new physics signalswhose main signature is missing ET. The addition of sen-sitive detector layers just outside the magnet (inside thefirst barrel muon station) was simulated and shown to en-sure much smaller energy leakage and hence improvementin the energy resolution.

Sensitive detector layers of 1 cm thick plastic scintilla-tor with embedded WLS fibre readout (similar technologyas the CMS barrel and endcap hadron calorimeters) wereselected as the appropriate detectors.

Signal efficiency0 0.2 0.4 0.6 0.8 1

Effic

ienc

y of

reje

ctin

g pe

dest

al

0

0.2

0.4

0.6

0.8

1

1.2

Fig. 19. Efficiency of rejecting pedestal vs. efficiency of muondetection in the HO detector. Different markers refer to differ-ent read out towers in ring 0.

6 Acknowledgement

The results presented in this paper are partially based onthe doctoral thesis of Seema Sharma[8]. This project wascarried out with financial support from CERN, Depart-ment of Atomic Energy and Department of Science andTechnology of India, U.S. Department of Energy, U.S. Na-tional Science Foundation, RMKI-KFKI (Hungary, OTKAgrant T 016823), Federal Agency for Science and Innova-tions of the Ministry for Education and Science of the Rus-sian Federation, Russian Academy of Sciences, Scientificand Technical Research Council of Turkey (TUBITAK),Turkish Atomic Energy Agency (TAEK), Bogazici Uni-versity Research Grant (Grant no: 04B301).

References

1. CMS Technical Proposal, CERN/LHCC 94-38, LHCC/P1,December 15, 1994.

2. CMS Hadron Calorimeter Technical Design Report,CERN/LHCC 97-31, CMS TDR 2, June 20, 1997.

3. CMS-HCAL Collaboration – V. V. Abramov et al ., NuclearInstruments and Methods A457 (2001) 75.

4. Sudeshna Banerjee and Sunanda Banerjee, CMS Note1999/063.

5. Outer Hadron Calorimeter Engineering Design Review,TIFR/CMS-99-01.

6. E. Hazen et al ., Nuclear Instruments and Methods A511

(2003) 311.7. C. Albajar et al ., CMS Note 2003/009.8. Seema Sharma,Ph.D. Thesis, Tata Institute of Fundamental

Research, Mumbai (unpublished), 2008.