performance of a large-area gem detector prototype for the upgrade of the cms muon endcap system
TRANSCRIPT
1
This work has been submitted to the IEEE Nuclear Science Symposium 2014 for publication in the conference record. Copyright may be transferred without notice, after which this version may no longer be available.
Performance of a Large-AreaGEM Detector Prototype for the
Upgrade of the CMS Muon Endcap SystemD. Abbaneo15, M. Abbas15, M. Abbrescia2, A.A. Abdelalim8, M. Abi Akl13, W. Ahmed8, W. Ahmed17,
P. Altieri2, R. Aly8, C. Asawatangtrakuldee3, A. Ashfaq17, P. Aspell15, Y. Assran7, I. Awan17, S. Bally15,Y. Ban3, S. Banerjee19, P. Barria5, L. Benussi14, V. Bhopatkar22∗, Member, IEEE, S. Bianco14, J. Bos15,
O. Bouhali13, S. Braibant4, S. Buontempo24, C. Calabria2, M. Caponero14, C. Caputo2, F. Cassese24,A. Castaneda13, S. Cauwenbergh16, F.R. Cavallo4, A. Celik9, M. Choi31, K. Choi31, S. Choi29, J. Christiansen15,
A. Cimmino16, S. Colafranceschi15, A. Colaleo2, A. Conde Garcia15, M.M. Dabrowski15, G. De Lentdecker5,R. De Oliveira15, G. de Robertis2, S. Dildick9,16, B. Dorney15, W. Elmetenawee8, G. Fabrice27, M. Ferrini14,
S. Ferry15, P. Giacomelli4, J. Gilmore9, L. Guiducci4, A. Gutierrez12, R.M. Hadjiiska28, A. Hassan8, J. Hauser21,K. Hoepfner1, M. Hohlmann22∗, Member, IEEE, H. Hoorani17, Y.G. Jeng18, T. Kamon9, P.E. Karchin12,
H.S. Kim18, S. Krutelyov9, A. Kumar11, J. Lee31, T. Lenzi5, L. Litov28, F. Loddo2, T. Maerschalk5,G. Magazzu26, M. Maggi2, Y. Maghrbi13, A. Magnani25, N. Majumdar19, P.K. Mal6, K. Mandal6,A. Marchioro15, A. Marinov15, J.A. Merlin15, A.K. Mohanty23, A. Mohapatra22, S. Muhammad17,
S. Mukhopadhyay19, M. Naimuddin11, S. Nuzzo2, E. Oliveri15, L.M. Pant23, P. Paolucci24, I. Park31,G. Passeggio24, B. Pavlov28, B. Philipps1, M. Phipps22, D. Piccolo14, H. Postema15, G. Pugliese2, A. Puig
Baranac15, A. Radi7, R. Radogna2, G. Raffone14, S. Ramkrishna11, A. Ranieri2, C. Riccardi25, A. Rodrigues15,L. Ropelewski15, S. RoyChowdhury19, M.S. Ryu18, G. Ryu31, A. Safonov9, A. Sakharov10, S. Salva16,
G. Saviano14, A. Sharma15, Senior Member, IEEE, S.K. Swain6, J.P. Talvitie15,20, C. Tamma2, A. Tatarinov9,N. Turini26, T. Tuuva20, J. Twigger22, M. Tytgat16, Member, IEEE, I. Vai25, M. van Stenis15, R. Venditi2,
E. Verhagen5, P. Verwilligen2, P. Vitulo25, D. Wang3, M. Wang3, U. Yang30, Y. Yang5, R. Yonamine5,N. Zaganidis16, F. Zenoni5, A. Zhang22
Manuscript received November 30, 2014.1RWTH Aachen University, III Physikalisches Institut A, Aachen, Germany2Politecnico di Bari, Universita di Bari and INFN Sezione di Bari, Bari,
Italy3Peking University, Beijing, China4University and INFN Bologna, Bologna, Italy5Universite Libre de Bruxelles, Brussels, Belgium6National Institute of Science Education and Research, Bhubaneswar, India7Academy of Scientific Research and Technology, ENHEP, Cairo, Egypt8Helwan University & CTP, Cairo, Egypt9Texas A&M University, College Station, USA10Kyungpook National University, Daegu, Korea11University of Delhi, Delhi, India12Wayne State University, Detroit, USA13Texas A&M University at Qatar, Doha, Qatar14Laboratori Nazionali di Frascati - INFN, Frascati, Italy15CERN, Geneva, Switzerland16Ghent University, Dept. of Physics and Astronomy, Ghent, Belgium17National Center for Physics, Quaid-i-Azam University Campus, Islam-
abad, Pakistan18Chonbuk National University, Jeonju, Korea19Saha Institute of Nuclear Physics, Kolkata, India20Lappeenranta University of Technology, Lappeenranta, Finland21University of California, Los Angeles, USA22Florida Institute of Technology, Melbourne, USA23Bhabha Atomic Research Centre, Mumbai, India24INFN Napoli, Napoli, Italy25INFN Pavia and University of Pavia, Pavia, Italy26INFN Sezione di Pisa, Pisa, Italy27IRFU CEA-Saclay, Saclay, France
Abstract—Gas Electron Multiplier (GEM) technology is beingconsidered for the forward muon upgrade of the CMS experimentin Phase 2 of the CERN LHC. Its first implementation is plannedfor the GE1/1 system in the 1.5 <| η |< 2.2 region of the muonendcap mainly to control muon level-1 trigger rates after thesecond long LHC shutdown. A GE1/1 triple-GEM detector isread out by 3,072 radial strips with 455 µrad pitch arrangedin eight η-sectors. We assembled a full-size GE1/1 prototypeof 1m length at Florida Tech and tested it in 20-120 GeVhadron beams at Fermilab using Ar/CO2 70:30 and the RD51scalable readout system. Four small GEM detectors with 2-Dreadout and an average measured azimuthal resolution of 36 µradprovided precise reference tracks. Construction of this largestGEM detector built to-date is described. Strip cluster parameters,detection efficiency, and spatial resolution are studied withposition and high voltage scans. The plateau detection efficiencyis [97.1 ± 0.2 (stat)]%. The azimuthal resolution is found to be[123.5 ± 1.6 (stat)] µrad when operating in the center of theefficiency plateau and using full pulse height information. Theresolution can be slightly improved by ∼ 10 µrad when correctingfor the bias due to discrete readout strips. The CMS upgradedesign calls for readout electronics with binary hit output.When strip clusters are formed correspondingly without charge-
28Sofia University, Sofia, Bulgaria29Korea University, Seoul, Korea30Seoul National University, Seoul, Korea31University of Seoul, Seoul, Korea∗Corresponding authors: [email protected], [email protected]
arX
iv:1
412.
0228
v1 [
phys
ics.
ins-
det]
30
Nov
201
4
weighting and with fixed hit thresholds, a position resolution of[136.8 ± 2.5 stat] µrad is measured, consistent with the expectedresolution of strip-pitch/
√12 = 131.3 µrad. Other η-sectors of the
detector show similar response and performance.
I. INTRODUCTION
DURING the phase II upgrade of the Compact MuonSolenoid (CMS) experiment at the CERN, CMS GEM
collaboration intends to install large-area GEM[1] detectors inthe forward muon endcap in the high-η region 1.5 <| η |< 2.2.Fig. 1 shows the quadrant of the muon system, where in-stallation of GE1/1 detectors is proposed. This installationwill help to restore redundancy for tracking and triggeringin the muon system, as GEM detectors provide very precisetracking information due to high spatial resolution. They canalso sustain high particles rates up to ∼MHz/cm2. CathodeStrips chambers (CSC) alone many times misidentify multiplyscattered lower pT muons as high pT muons. This problemcan be overcome by using GEM detectors in conjuction withthe CSC system as shown in Fig. 2; together they providean accurate measurement of the muon bending angle that isnot affected by multiple scattering. This discriminates lowerpT muons from higher pT muons and reduces the soft muonrate at the level-1 trigger as shown in Fig. 3, which will helpcontrol the muon trigger rate at the high luminosity LHC.
0 2 4 6 8 10 12 z (m)
R (m
)
1
0
2
3
4
5
6
7
8
1 3 5 7 9 115.0
4.0
3.0
2.52.42.32.22.12.0
1.9
1.8
1.7
1.6
1.5
1.4
1.3
1.2
1.00.9 1.10.80.70.60.50.40.30.20.140.4°44.3° 36.8°48.4°52.8°57.5°62.5°67.7°73.1°78.6°84.3°
0.77°
2.1°
5.7°
9.4°10.4°11.5°12.6°14.0°15.4°
17.0°
18.8°
20.7°
22.8°
25.2°
27.7°
30.5°
33.5°
θ°η
θ°η
ME4
/1
ME3
/1
ME2
/1
ME1
/2
ME1
/1
ME2
/2
ME3
/2
ME1
/3
RE3
/3
RE1
/3R
E1/2
MB1
MB2
MB3
MB4
Wheel 0 Wheel 1
RB1
RB2
RB3
RB4
HCAL
ECAL
Solenoid magnet
Silicon tracker
Steel
RE2
/2
Wheel 2
RE2
/3
RE3
/2M
E4/2
RE4
/3R
E4/2
GE1
/1
DTsCSCsRPCsGEMs
Fig. 1. A quadrant of the CMS muon system with proposed GE1/1 upgradein red.
Fig. 2. GEM and CSC system in each station enlarge the lever arm for abending angle measurement unaffected by multiple scattering.
(a)
(GEM,CSC) [rad]φ∆0 0.005 0.01 0.015 0.02 0.025 0.03
Arb
itray
uni
ts
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
0.016
0.018 GEM-CSC Bending Angle CMS Phase-2 Simulation Preliminary
= 5 GeV/cT
muon p = 20 GeV/c
Tmuon p
|<2.14η1.64<|Odd ("far") chamber pairs
GEM-CSC Bending Angle CMS Phase-2 Simulation Preliminary
(b)
2 3 4 5 6 7 8 910 20 30 40 50 210
Trig
ger r
ate
[kHz
]-210
-110
1
10
210
310
410 CMS Phase-2 Simulation Preliminary
L1 Single Mu (2012 configuration) + ME1/1a unganging3 stubs≥GEM+CSC integrated trigger with
-1 s-2 cm34L = 4*10
| < 2.14η1.64 < |
CMS Phase-2 Simulation Preliminary
[GeV/c]cutT
p2 3 4 5 6 7 8 910 20 30 40 50 210
Ratio
-210
-110
1
(GEM+CSC)/L1 Single Mu
Fig. 3. (a) Simulation of the bending angle measurement in the first endcapstation (GE1/1-ME1/1) for soft (∼5 GeV) and hard (∼20 GeV) muons[2].(b) Simulation of the inclusive muon trigger rate expected for the LHC Phase2 as function of the Level-1 pT trigger threshold in 1.5 <| η |< 2.2.
We have constructed a 1m-long GE1/1 prototype, one ofthe largest GEM detectors in existance, at Florida Institute ofTechnology and tested it in a hadron beam at Fermilab. Thisparticular prototype represents the third generation of GE1/1prototypes. Performance characteristics of the detector such asstrip cluster parameters, detection efficiency, and spatial res-olution have been studied using full pulse height informationas well as with binary hit reconstruction. A correction for thenon-linear strip response is also implemented in an attempt toimprove overall spatial resolution of the GE1/1 detector.
II. CONSTRUCTION OF A LARGE-AREA GEM DETECTORAT FLORIDA TECH
The large-area GEM detector proposed for the CMS muonendcap upgrade is a trapezoidal Triple-GEM detector withapproximately 99×(28-45) cm2 active surface area. This de-tector has a 3/1/2/1 mm (drift, transfer 1, transfer 2, induction)internal electrode gap configuration. The GEM foils used inthis detector are produced by single-mask etching technique atCERN. Each GEM foil is divided into 35 high voltage sectorsthat are transverse to the direction of the readout strips. A stackof these three GEM foils is mounted on the drift electrode; the
2
gas volume is closed with the readout board. This detector isconstructed using an internal mechanical stretching method[3]introduced in 2011. There are a total of 3072 radial readoutstrips with 455 µrad pitch along the azimuthal direction anddistributed over eight η-sectors. In each of the sectors, inducedsignals are read out via 384 radial strips. Fig. 4 shows the mainsteps involved in the construction of the GE1/1-III prototypedetector.
(a)
Step IGEM foil stack assembly
Step IIStretching GEM foils by providing tension
(b)
Step III Finished GEM detector w/ APV frontend hybrids connected
1D readout board with 3,072 radial strips connected
through 24 Panasonic connectors to 24 APV hybrids
Eta 8 7 6 5 4 3 2 1
Fig. 4. Construction of the 1m-long large-area GE1/1-III prototype in threesteps. The instrumentation with APV front-end hybrids and the numbering ofeta sectors is shown in (b).
III. TEST BEAM SETUP
The GE1/1-III prototype detector was tested in a 32 GeVmixed hadron beam at the Fermilab test beam facility(FTBF) in October 2013. For tracking studies, this de-tector was positioned on a movable table between fourGEM detectors with 2D Cartesian readout[4]. Three of themwere small 10 cm × 10 cm GEM detectors and one was a50 cm × 50 cm (Provided by University of Virginia) GEMdetector for which only the central 10 cm × 10 cm area wasinstrumented shown in Fig. 5. These 2D GEM detectors con-tained 256 straight strips along each horizontal (y-coordinate)and vertical (x-coordinate) plane with 0.4 mm pitch. The datawere collected using the RD51 scalable readout system[5][6]with an external trigger provided by scintillators. During thisbeam test, all detectors used an Ar/CO2 70:30 gas mixture.
Trackers
Trackers
CMS GE1/1-III Prototype detector
Fig. 5. Experimental setup in the beam line MT6.2-B at FTBF.
IV. BEAM TEST RESULTS
A. Performance Characteristics
Performance characteristics of the GE1/1 detector werestudied using high voltage and position scan data. In the highvoltage scan, the beam was focused on sector 5 and the voltageof the GE1/1 detector was varied from 2900V to 3350V,whereas voltages of all tracker detectors were set so that thetrackers operated on their efficiency plateaus throughout allmeasurements. Fig. 6 shows the cluster charge distributionfor η-sector 5 at 3250V. All distributions obtained at differenthigh voltages are fitted with a Landau function and the MostProbable Values (MPV) obtained from these fits are used forobtaining the uniformity results for the GE1/1 detector.
Fig. 6. GE1/1 cluster charge distribution at 3250V fitted with Landau function.
3
The number of strips above threshold in a strip cluster definethe strip multiplicity of that cluster. For the GE1/1 detector, thedistribution of multiplicity in strip clusters is shown in Fig. 7;the average strip multiplicity is 2.4 strips. Fig. 8 shows thatthe strip multiplicity increases with high voltage, i.e. gas gain.
Fig. 7. Strip multiplicity in GE1/1 clusters at 3250V.
Fig. 8. Mean strip multiplicity increases with high voltage.
The detection efficiency is obtained as ε = N1
N−N2, where
N = total number of triggered events; N1 = number ofevents with Cluster Multiplicity (CM) ≥ 1 for given sector;N2 = number of events with CM ≥ 1 for other sectors. Afew strips from neighboring sectors triggered sometimes dueto scattering of particles or because the beam was hittingnear the edge of the sector. To obtain accurate efficiencyof the given sector, we eliminate such events from the totalnumber of events. The overall detection efficiency of the
detector is measured for four threshold cuts on the pedestalwidth, namely 3σ, 4σ, 5σ, and 6σ, where σ is the width ofthe pedestal distribution from a Gaussian fit. Fig. 9 showsefficiency curves fitted with a sigmoid function for thesethreshold cuts. The efficiency curves show a long plateaufor higher voltages and the efficiency values on plateau arenot much affected by different threshold cuts. The detectionefficiency of the GE1/1 detector with 5σ cut on the pedestalwidth is [97.1 ± 0.2 (stat)]%.
Drift-HV2850 2900 2950 3000 3050 3100 3150 3200 3250 3300 3350
Eff
icie
ncy
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
A]µHV Divider Current [660 670 680 690 700 710 720 730 740 750 760
0.0025±Sigma3 eff: 0.9708
0.0022±Sigma4 eff: 0.9677
0.002±Sigma5 eff: 0.978
0.002±Sigma6 eff: 0.9623
1 + exp(p0 * (x + p1))p2
Fig. 9. Detection efficiency with different strip threshold cut on pedestalwidth.
Fig. 10. Response uniformity in seven η-sectors for three different APVpositions, i.e. Upper, Middle, and Lower at 3250V.
The position scan is used to measure the uniformity re-sponse of the detector for all η-sectors at operating voltage3250V. The scan was performed for three APV positions(see Fig. 4 (b)). For each sector, the charge uniformity wasmeasured using the Landau MPV from the cluster charge
4
distribution. Fig. 10 shows the uniformity response for the firstseven sectors. The uniformity varies by ∼25% across differentsectors. The variation in sector 6 and 7 has been improvedin subsequent prototypes by carefully stretching foils in thatregion and by minimizing any bending of pcbs.
The response uniformity of the detector is also calculatedby considering only single-strip single-cluster events. Fig. 11shows the uniformity response of the η-sectors of the detector.
Fig. 11. Charge uniformity for different η sectors at 3250V using single-stripsingle-cluster events.
B. Tracking Analysis
Beam profiles can be obtained using 2D hit maps of thereference tracker detectors. The beam profile for the secondary32 GeV mixed-hadron beam is oval while the beam profile forthe 120 GeV proton beam is more circular and more focusedas shown in Fig. 12.
(a) (b)
Fig. 12. Beam profiles: (a) 32 GeV hadron beam. (b) 120 GeV proton beam.
The tracking analysis is done in three steps to study thespatial resolution of the GE1/1 detector. The first step isalignment which is performed in two steps. The first oneis to shift all tracker detectors so that their origins in x-ycoordinate coincide and their residuals are centered at zero.In the second step, the three downstream tracking detectorsare rotated with respect to the first tracker detector by usinginitial shift parameters from step one. The rotating angle foreach detector is optimized such that the residual width of eachdetector is minimized.
The next step in the tracking analysis is atransformation from Cartesian (x, y) coordinates topolar (r, φ) coordinates, because the GE1/1 detector measuresthe azimuthal coordinate (φ) with its radial strips. Fig. 13shows correlation in φ for hits in the GE1/1 detector and inthe first tracker detector.
Fig. 13. Correlation of GE1/1 detector hits with hits in first tracker detector.
(a)
(b)
Fig. 14. Residuals for tracker 1 in ϕ : (a) Inclusive residual with widthσ = 21 µrad and (b) Exclusive residual with width σ = 75 µrad.
5
Fig. 15. Residuals and angular resolution of the tracking detectors
In the final step of the tracking analysis, both inclusive andexclusive track-hit residuals are calculated. The definition ofan inclusive (exclusive) residual is the residual calculated byincluding (excluding) the probed detector in the track fitting.The spatial resolution of the detector is calculated from thegeometric mean[7] of the widths of inclusive and exclusiveresiduals, i.e. σ =
√σinc × σexc[8]. Fig. 14 shows both
residual widths for the first tracker detector. The inclusiveresidual width is σinc = 21 µrad and exclusive residual widthis σexc = 75 µrad; with a geometric mean of ∼40 µrad. Sim-ilarly, the residual widths of the other three tracker detectorsare calculated. Fig. 15 summarizes the angular resolution ofall four tracking detectors.
The angular resolution of the GE1/1-III prototype GEMdetector is calculated using two different methods forobtaining the hit position. The barycentric method uses thefull pulse height information to find the strip cluster barycenterand the binary method uses binary hits reconstructed offlinefrom the analog APV readout to emulate the behavior ofVFAT[9] chips with binary output. The results are as follows:
1) The Barycentric Method: Fig. 16 shows inclusive andexclusive residuals of the GE1/1 detector at the center ofsector 5 at 3250V using a 5σ cut on the pedestal width. Theinclusive residual is 110.7 µrad and the exclusive residualis 137.9 µrad in azimuthal direction; this corresponds to208.1 µm and 259.3 µm, respectively, with R = 1880.5 mm.The resulting resolution of the GE1/1 detector using pulseheight analog readout is [123.3 ± 1.6 (stat)] µrad, i.e. 27% ofthe strip pitch. Residuals for sector 5 are plotted against thedifferent applied high voltages in Fig. 17. The best resolutionis obtained on the efficiency plateau as expected.
2) The Binary Method (Emulated VFAT Threshold): Duringthis beam test, all data were collected using APV25 hybridchips that provide the analog signal, whereas for the CMSupgrade electronics the VFAT3 is proposed that produces abinary output for each readout strip (charge above or belowthreshold). Hence, it is important to study the characteristics
(a)
residualEta5_phi_X_-1880.5_Y_-28.7_angle_-0.000004
Entries 6447
Mean 3.393e-006
RMS 0.0001887
/ ndf 2χ 179.4 / 68
Prob 5.577e-012
Primary Constant 13.3± 867.3
Primary Mean 1.500e-006± -1.958e-006
Primary Sigma 0.0000010± 0.0001107
Background Constant 1.194± 5.539
Background Mean 0.0000582± 0.0001477
Background Sigma 0.0001168± 0.0007292
[rad]φResidual in -0.002 -0.0015 -0.001 -0.0005 0 0.0005 0.001 0.0015 0.002
Fre
qu
en
cy
0
100
200
300
400
500
600
700
800
900residualEta5_phi_X_-1880.5_Y_-28.7_angle_-0.000004
Entries 6447
Mean 3.393e-006
RMS 0.0001887
/ ndf 2χ 179.4 / 68
Prob 5.577e-012
Primary Constant 13.3± 867.3
Primary Mean 1.500e-006± -1.958e-006
Primary Sigma 0.0000010± 0.0001107
Background Constant 1.194± 5.539
Background Mean 0.0000582± 0.0001477
Background Sigma 0.0001168± 0.0007292
number of strips-4 -3 -2 -1 0 1 2 3 4
(b)
residualEta5_phi_X_-1880.5_Y_-28.7_angle_-0.000004
Entries 6447
Mean 2.499e-006
RMS 0.00022
/ ndf 2χ 189.3 / 76
Prob 1.208e-011
Primary Constant 10.6± 692.5
Primary Mean 1.89e-006± -1.48e-006
Primary Sigma 0.0000013± 0.0001379
Background Constant 0.83± 4.96
Background Mean 0.000071± 0.000153
Background Sigma 0.0001037± 0.0008787
[rad]φResidual in -0.002 -0.0015 -0.001 -0.0005 0 0.0005 0.001 0.0015 0.002
Fre
qu
en
cy
0
100
200
300
400
500
600
700residualEta5_phi_X_-1880.5_Y_-28.7_angle_-0.000004
Entries 6447
Mean 2.499e-006
RMS 0.00022
/ ndf 2χ 189.3 / 76
Prob 1.208e-011
Primary Constant 10.6± 692.5
Primary Mean 1.89e-006± -1.48e-006
Primary Sigma 0.0000013± 0.0001379
Background Constant 0.83± 4.96
Background Mean 0.000071± 0.000153
Background Sigma 0.0001037± 0.0008787
number of strips-4 -3 -2 -1 0 1 2 3 4
Fig. 16. (a) Inclusive GE1/1 residual with σ = 110.7 µrad and (b) ExclusiveGE1/1 residual with σ = 137.9 µrad using full pulse height information anddouble-Gaussian fits.
HV [V]2900 3000 3100 3200 3300
rad]
µR
esol
utio
n [
110
120
130
140
150
160
170Exclusive residual width
Inclusive residual width
Resolution (geometric mean)
Fig. 17. Resolutions of the GE1/1 detector at the center of sector 5 fordifferent voltages using full pulse height information.
of the GE1/1 detector using binary output. By applying afixed threshold cut on the pedestal width, binary hits arereconstructed offline from the pulse height data. The detectorefficiency is calculated again by applying fixed cuts of 0.8 fC,0.98 fC, and 1.2 fC, which are equivalent to 10 VFAT,
6
12 VFAT, and 15 VFAT units; where 1 VFAT unit = 0.08 fC.The efficiency curve is again fitted with the sigmoid functionand from its parameter the efficiency of the detector is foundto be [96.9 ± 0.2 (stat)]% on the plateau.
Fig. 18. Detection efficiency using VFAT-like binary hit data with threedifferent thresholds.
(a)
residualEta5_phi_X_-1858.2_Y_-29.0_angle_-0.000107
Entries 5052
Mean -9.383e-005
RMS 0.0002128
/ ndf 2χ 46.24 / 71
Prob 0.99
Primary Constant 11.2± 603.5
Primary Mean 0.0000019± -0.0001012
Primary Sigma 0.0000016± 0.0001228
Background Constant 1.129± 5.897
Background Mean 5.654e-005± 2.421e-005
Background Sigma 0.0001061± 0.0007549
[rad]φResidual in -0.002 -0.0015 -0.001 -0.0005 0 0.0005 0.001 0.0015 0.002
Fre
quen
cy
0
100
200
300
400
500
600residualEta5_phi_X_-1858.2_Y_-29.0_angle_-0.000107
Entries 5052
Mean -9.383e-005
RMS 0.0002128
/ ndf 2χ 46.24 / 71
Prob 0.99
Primary Constant 11.2± 603.5
Primary Mean 0.0000019± -0.0001012
Primary Sigma 0.0000016± 0.0001228
Background Constant 1.129± 5.897
Background Mean 5.654e-005± 2.421e-005
Background Sigma 0.0001061± 0.0007549
number of strips-4 -3 -2 -1 0 1 2 3 4
(b)
residualEta5_phi_X_-1858.2_Y_-29.2_angle_-0.000107
Entries 5052
Mean -2.114e-005
RMS 0.0002475
/ ndf 2χ 58.51 / 77
Prob 0.9423
Primary Constant 9.0± 482.7
Primary Mean 2.356e-006± -3.158e-005
Primary Sigma 0.0000020± 0.0001525
Background Constant 1.104± 5.568
Background Mean 6.11e-005± 9.35e-005
Background Sigma 0.0001103± 0.0008229
[rad]φResidual in -0.002 -0.0015 -0.001 -0.0005 0 0.0005 0.001 0.0015 0.002
Fre
quen
cy
0
100
200
300
400
500 residualEta5_phi_X_-1858.2_Y_-29.2_angle_-0.000107
Entries 5052
Mean -2.114e-005
RMS 0.0002475
/ ndf 2χ 58.51 / 77
Prob 0.9423
Primary Constant 9.0± 482.7
Primary Mean 2.356e-006± -3.158e-005
Primary Sigma 0.0000020± 0.0001525
Background Constant 1.104± 5.568
Background Mean 6.11e-005± 9.35e-005
Background Sigma 0.0001103± 0.0008229
number of strips-4 -3 -2 -1 0 1 2 3 4
Fig. 19. (a) Inclusive GE1/1 residual with σ = 122.8 µrad and (b) ExclusiveGE1/1 residual with σ = 152.5 µrad using binary hit reconstruction anddouble-Gaussian fits.
Fig. 19 (a) shows that the inclusive residual width ofthe detector is 122.8 µrad (228.1 µm) and (b) shows
that the exclusive residual width is 152.5 µrad (283.3 µm)with binary hits. The geometric mean of these widths is136.8 µrad (254.2 µm), which is consistent with the expec-tation from angular strip pitch/
√12 = 131.3 µrad.
The total radiation length of all chambers used in the testbeam setup is about ∼14% which causes multiple scatteringof the 32 GeV beam particles on the order of ∼147 µrad RMSof the multiple scattering angle. This value is not negligibleas it is within the range of the expected residual widths of thedetector. We are currently setting up a Monte Carlo simulationof the full tracking setup including material budgets of allchambers to correct for multiple scattering. We expect thatthe corrected result for the resolution will be somewhat betterthan the above preliminary results.
C. Correction of non-linear strip response
The motivation for implementing this correction for thebarycentric method is to further improve the spatial resolutionof the GE1/1 detector. In this study, the strip cluster positionis reconstructed in the detector via the cluster barycenter, orcentroid[10]. A η-correction factor is developed to correct thebarycenter position in the GE1/1 detector for biases due tothe discretized readout with strips. For the strip cluster withmore than one strip, η is defined as η = sb−smax[11], wheresb =
∑ni=1
qi·siqtotal
gives the barycenter position in terms ofstrip numbers; si and qi are strip number and charge of the ith
strip, respectively; smax is the strip number with the maximumcharge. This correction was done for 2, 3, and 4-strip clusters.Fig. 20 (a) and (b) show the scatter plots of exclusive residual
(a) (b)
(c) (d)
Fig. 20. (a) and (b) Scatter plots of exclusive residual vs. η for 2-strip and3-strip cluster, respectively, for combined HV Scan data. (c) and (d) Scatterplots of exclusive residual vs. η for all strip cluster multiplicities before andafter correction, respectively.
against η for 2-strip and 3-strip clusters, respectively. Theyshow different behavior and hence are fitted with differentfunctions. Finally, a corrected resolution is obtained for theGE1/1 detector after correcting the measured hit positionsby subtracting the value of the fitted function at η from
7
the original hit position. Fig. 20 (c) shows a scatter plot ofthe residuals for 2-strip and 3-strip clusters before correctingthe hit position and (d) shows the corrected residuals forthese strip clusters. For the high voltage scan data, the strip
(a)
HV [V]2900 3000 3100 3200 3300
rad]
µRe
solu
tion
[
105
110
115
120
125
130
135
140Correction with 2-strip clusters
Before correction
After correction
(b)
HV [V]3050 3100 3150 3200 3250 3300 3350
rad]
µRe
solu
tion
[
110
115
120
125
130
135
140
145
150
155
160 Correction with 3-strip clustersBefore correctionAfter correction
(c)
HV [V]2900 3000 3100 3200 3300
rad]
µRe
solu
tion
[
120
125
130
135
140
145
150
155All clusters, only corrected 2, 3 and 4-strip clusters
Before correction
After correction
Fig. 21. (a) Resolutions before and after correction for 2-strip cluster. (b)Resolutions before and after correction for 3-strip cluster. (c) Resolutionsbefore and after correction for 2-strip, 3-strip, and 4-strip cluster for highvoltage scan.
correction factor changes negligibly over the entire voltagerange. Consequently, all HV scan data were combined beforeimplementing the strip correction for more statistics. However,the strip correction was performed individually on each sectorfor the position scan. Fig. 21 shows resolutions before and
after correction. Since the readout of the GE1/1 has a finestrip segmentation, the overall improvement factor is small(less than 8%, i.e. within ∼ 10 µrad) compared to detectorshaving coarser readout strips[12].
V. SUMMARY AND CONCLUSION
A CMS GE1/1-III prototype GEM detector was successfullybuilt and tested at Florida Tech and Fermilab, respectively,in 2013. The detector performed well in terms of detectorefficiency, which is greater than 97% with two differentmethods for defining hits. It shows good charge uniformityfor all η-sectors except for sectors 6 and 7. The uniformityshould be improved in future assemblies by making sure thatall the foils have the same tension in all eight sectors. Thespatial resolution of the GE1/1 detector is ∼123 µrads with thebarycentric method and ∼136 µrads with the binary method.The binary method meets the expected value of resolutionfrom the pitch of the strip. The spatial resolution of thebarycentric mehod of the detector is improved by ∼10 µradafter correcting for non-linear strip response. In conclusion, thetested GE1/1 prototype detector meets all but one performanceexpectation for use in the CMS muon endcap upgrade.
VI. ACKNOWLEDGEMENT
We thank the RD51 collaboration for its technical support.Also we thank the FLYSUB consortium as well as the staffof the Fermilab Test Beam Facility for their support duringthe beam test. Also, we gratefully acknowledge support fromFRS-FNRS (Belgium), FWO-Flanders (Belgium), BSF-MES(Bulgaria), BMBF (Germany), DAE (India), DST (India),INFN (Italy), Uni. Roma (Italy), NRF (Korea), QNRF (Qatar),DOE (USA), and WSU (USA).
REFERENCES
[1] F. Sauli, ”GEM: A new concept for electron amplification in gas detec-tors,” Nucl. Instrum. Meth. A 386, 531 (1997).
[2] M. Tytgat et al., ”Status of the Triple-GEM project for the up-grade of the CMS Muon System,” IOP science, 2013 Avail-able: http://www.ieee.org/portal/cms docs/pubs/confstandards/pdfs/IEEE-PDF-SpecV401.pdf.
[3] S. Colafranceschi, M. Hohlmann, et al., ”Beam Test Results for NewFull-scale GEM Prototypes for a Future Upgrade of the CMS High-ηMuon System,” in Proc. IEEE NSS-MIC 2012, N14-137, pp. 1172-1176,arXiv:1211.3939v1 [physics.ins-det].
[4] B. Ketzer, et al., ”Performance of triple GEM tracking detectors in theCOMPASS experiment,” Nucl. Instrum. Meth. A 535 (2004) p.314-318.
[5] K. Gnanvo, et al., ”Detection and Imaging of High-Z Materials with aMuon Tomography Station Using GEM Detectors,” in Proc. IEEE NSS-MIC 2010, NI19-90, pp. 552-559, arXiv:1011.3231 [physics.ins-det].
[6] ALICE DAQ and ECS Manual, 2010, ALICE Internal Note/DAQ,ALICE-INT-2010-001.
[7] R.K. Carnegie et al., ”Resolution studies of cosmic-ray tracks in a TPCwith GEM readout,” Nucl. Instrum. Meth. A, 535(2005), p. 372-383.
[8] D.C. Arogancia et al., Nucl. Instrum. Meth. A, 602(2007), p. 403-414.[9] P. Aspell, et al., ”The VFAT production test platform for the TOTEM
experiment,” in Proc. Topical Workshop on Electronics for ParticlePhysics 2008, CERN-2008-008.
[10] Gregorio Landi, ”Properties of the center of gravity as an algorithmfor position measurements,” Nucl. Instrum. and Meth. A, 485 (2002) p.698-719.
[11] Marco Villa, CERN-Thesis-2013-284.[12] A. Zhang, et al., ”Study of a large-area GEM detector readout with radial
zigzag strips for forward tracking in experimentas at a future Electron-IonCollider,” IEEE NSS-MIC 2014, N05-12.
8