ATL-GEN-SLIDE-2009-356 18 November 2009 The Status of ATLAS Xin Wu, University of Geneva On behalf of the ATLAS collaboration 1
ATLAS and the people who built it 25m high, 44m long Total weight 7000 tons 10 8 electronic channels 3000 km of cables ~2900 physicists (~1050 PhD students) Many engineers and technicians 172 Institutions from 37 countries and 5 continents Oct. 2009 numbers 2
20 years and the end of the beginning 1989: R&D starts 1992 : ATLAS LoI 1994: TP 1996: approval From 1996 on a series of TDR 1997: construction starts 2003: underground installation starts 2008: installation completed Sept. 2008: single-beam events recorded Dec. 2009: first LHC collision recorded? Closure of the LHC beam pipe inside the ATLAS cavern on 16 June 2008. First beam splash event,10 Sep. 2008 3
Detector hardware status summary in one page Sub-detector Number of channels Operational fraction (%) Pixels SCT Silicon Strips TRT Transition Radiation Tracker 80 M 6.3 M 350 k 98.0 99.3 98.2 LAr EM Calorimeter Tile Calorimeter Hadronic Endcap LAr Calorimeter Forward LAr Calorimeter 170 k 9.8 k 5.6 k 3.5 k 98.8 99.5 99.9 100 MDT Muon Drift Tubes CSC Cathode Strip Chambers RPC Barrel Muon Trigger Chambers TGC Endcap Muon Trigger Chambers 350 k 31 k 370 k 320 k 99.7 98.4 97.0 99.8 Trigger and DAQ: routinely taking long runs of cosmic data with all detector integrated at >300 MB/s 4
More detailed report on ATLAS Status Magnet System Muon Spectrometer Calorimeters Inner Detector Trigger and DAQ Computing Combined Performance with Cosmics 5
Magnet System 6
Magnet System Barrel toroid 8 superconducting coils, each 25 m long, 5m wide, 100 tons I=20.5 ka, T=4.5 K Typical field 0.5 T Endcap toroid (x2) 8 coils in common cryostat 11m diameter, 240 tons I=20.5 ka, T=4.5 K Typical field 1T Solenoid 5.8m long, 2.5m diameter. I=7.7 ka, T=4.5 K Field 2 T Closure of Endcap toroid in May 2009 Insertion of solenoid in LAr Calo in Feb 2004 7
Full barrel toroid in place in October 2005 8
Operation of the magnet system The full magnet system has been operated at full current for long periods since August 2008 June - July 2009 9
Muon Spectrometer MDT RPC TGC CSC 10
Barrel Muon Spectrometer Precision chambers Trigger chambers 700 barrel precision chambers (MDT) 600 barrel trigger chambers (RPC, η <1.05) Momentum resolution: <10% up to ~1 TeV Field integral: 2-6 Tm η <1.3, 4-8 Tm 1.6< η <2.7 11
Forward Muon Spectrometer Big wheels (and end-wall wheels): Small wheels : ~400 MDT precision chambers ~32 CSC precision chambers ~3600 TGC trigger chambers ~80 MDT chambers Small wheels End-wall wheels Big wheels 12
Muon System Hardware Status MDT Sub-detector Number of channels Operational fraction MDT CSC RPC TGC 341 k 31 k 359 k 318 k 99.7% 98.4% 97.0% 99.8% RPC 13
Triggering on Cosmic Muons Both RPC and TGC have been providing stable cosmic muon triggers since August 2008 Access shafts Elevators 14
Muon Detection Efficiency TGC layer efficiency ~94% as designed TGC preliminary phi sector MDT segment efficiency ~99% as designed MDT RPC Barrel middle chambers RPC layer efficiency ~94% will be improved to ~97% designed eta sector 15
Muon Detector Alignment with Cosmics Unbiased sagitta of straight tracks for MDT middle barrel chamber BML2C03 Nominal geometry Alignment procedure has been tested with cosmic muons With tracks from collisions, will align to ~30 µm of accuracy, to achieve 10% p/p at 1 TeV. B~0.5T sagitta ~ 500 µm at 1 TeV With optical alignment y L~5m z Details in O. Kortner s talk this afternoon After track alignment 2008 cosmic data 16 12,232 optical sensors
Muon Detector Tracking Performance Standalone tracking efficiency standalone momentum resolution with cosmics Efficiency calculated with respect to Inner Detector tracks Degradation due to limited alignment statistics and cosmic timing jitter Details in O. Kortner s 17 talk this afternoon
Calorimeters Fe/Scint Tile LAr 18
Tile Hadronic Calorimeter Sampling calorimeter with Fe/scintillator (4.7:1), 7.4λ, covering η <1.7 Projective geometry cells with η ϕ=0.1x0.1 (0.2x0.1 in third layer) Double PMT readout via WLS fibers (5000 cells 10k channels) Expected jet energy resolution E E ~ 50% E 3% 19
Tile Calorimeter Noise Stability Noise stability Aug-Nov 08 RMS (10k channels)= 0.31 % Mean ~ 1.44 ADC counts Noise remains stable over months as measured in random events Minimum ionizing particle signal from cosmic muons cleanly above noise 20
Tile Calorimeter Uniformity Cells in second layer phi Cell response in eta and phi extrapolated from the Inner Detector track Cells in first layer eta Good response uniformity and ID-TileCal alignment 21
Liquid Argon Calorimeter Electromagnetic: Barrel + End-cap ( η <3.2) Pb+LAr accordion, ~22-35 X 0 σ/e ~ 10%/ E 0.7% Hadronic: End-cap (1.5< η <3.2) Cu+LAr flat-plate, 11 λ σ/e ~ 50%/ E 3% Forward: (3.1< η <4.9) EM: Cu+lAr, HAD: W+LAr small LAr gaps between rods and tubes σ/e ~ 100%/ E 10% 22
LAr Hardware Status ~182k channels in total LAr HV correction factors: ~94% with nominal values 0.02% irrecoverable ~1% with dead readout, mainly due to bad optical transmitters on front end boards, to be fixed at next access 23
LAr Noise and Stability Electronics noise in MeV Etmiss in random triggers Pedestal stability Gain stability 24
LAr Pulse Shape and Drift Time Measurement Pulse shapes are validated with cosmic data with 32 samples readout Agreement is better than 2% Drift time extracted from pulse shape measurement in good agreement with expectation Relation between drift time and gap size allows to set limit on contribution to the energy resolution constant term In the barrel T dr /T dr =1.37±0.03% C gap < 0.32% (test beam ~0.,2%) E cell > 1 GeV 25
Energy Reconstruction and Uniformity of LAr S1 S2 Select projective tracks Associated energy reconstructed across the EM barrel in the first and second layers Measured non-uniformity in η sets a limit of 1.1% at 95% CL for the middle layer Details in T. Koffas talk Thursday afternoon 26
Inner Detector Silicon TRT 27
Inner Detector Pixel : 80 M 50x400µ pixels TRT : 350k straw tubes (Xe) SCT : 6.3 M 80µ Si strips 28
Inner Detector Tracking Tracking in η <2.5, B=2T A track in barrel passes: 36 TRT straws 4x2 silicon strips 3 pixels Expected performance: σ/p T ~ 3.4x10 4 p T (GeV) 0.015 σ d0 ~ 10 140/p T (GeV) µm 29
Noise Occupancy Very good noise occupancy Pixel: < 10 9 SCT: < 5 10 4 TRT: 2% SCT 10 10 Pixel 30
Hit Efficiency SCT 99.5% Very good hit efficiency Pixel, SCT: >>99% TRT: > 97% 99.7% TRT Pixel 31
Alignment with Cosmic Ray Tracks Details in L. Vacavant s talk Tuesday morning Pixel SCT TRT Results of alignment using 2008 cosmics Pixel: σ x perfect 16µ, achieved 24µ σ y perfect 127µ, achieved 131µ SCT: perfect 24µ, achieved 30µ TRT: perfect 136µ, achieved 165µ standalone and using full ID track in good agreement 32
Detector Mechanical Stability Pixel SCT Residual distribution of 2009 cosmic data using 2008 alignment Pixel x: 26µ, SCT : 36µ Detector is mechanically stable 33
Tracking Performance Details in L. Vacavant s talk Tuesday morning Fitting a cosmic ray passing through the full detector as two separate tracks (upper and lower) allows to measure the track resolutions directly Already close to ideal detector performance! Impact parameter resolution momentum resolution 34
Trigger and DAQ (TDAQ) High Level Trigger CPU Farm L1 Trigger racks HLT farm: 850 multi-core PCs installed (35% of the total planned for high luminosity) 35
TDAQ Architecture Level 1 (L1) Custom electronics identifies Region of Interest (RoI) using info from calorimeter and muon detectors High level Trigger (HLT) L2: PC farm runs fast algorithms to reconstruct e/γ, µ, τ, in a RoI using full granularity info from all detectors EF: PC farm runs offline algorithms accessing the full event 300 MB/s to mass storage ~3 PB/year stored DAQ Pipelined front end electronics Readout System with custom built buffers in PC farm Event building in PC farm on Data Network 36
TDAQ in full action since early 2008 3 major cosmic data taking (24/7) periods with fully integrated detectors ~200M events recorded in Sept/October 2008 ~90M event recorded in June/July 2009 global cosmic data taking with full detector resume in October 2009 L1 triggered by L1Muon, L1Calo, MinBias scintillator triggers HLT used to select tracks passing through Inner Detector many HLT algorithms are exercised 37
Trigger Performance with Cosmics Correlation between L1Calo and offline Track trigger efficiency at L2 L2 and EF shower shape for electron id comparison of EF and offline muon reconstruction 38
Computing The LHC computing universe 39
Computing Infrastructure ~ 50 PB/year of ATLAS data to be moved across the Grid 10 Tier-1s ~70 Tier-2s 10 9 raw events per year to be processed/reprocessed ~ 300 MB/s RAW DATA CERN Tier-O Hundreds of physicists analyzing data simultaneously CA Tier-1 DE Tier-1 ES Tier-1 FR Tier-1 IT Tier-1 NL Tier-1 NDGF Tier-1 TW Tier-1 UK Tier-1 US Tier-1 Tier-2s Tier-2s Tier-2s Tier-2s Tier-2s Tier-2s Tier-2s Tier-2s Tier-2s Tier-2s World-wide LHC Computing Grid 40
Computing Infrastructure Performance # of jobs 32k Oct 2008 - Oct 2009 >32k simultaneously running jobs reached Several reprocessing campaigns using 2008 and 2009 cosmics datasets 4GB/s June 2009 Higher than nominal data transfer (T0 to T1s and between T1s) sustained over 2 weeks (1-2GB/s at LHC). 41
Combined Performance with Cosmics hundreds of millions of these Highlights given here. More details in later talks: e, µ, τ : this afternoon by O. Kortner Tracking and b-tag: Tuesday morning by L. Vacavant Jet and Etmiss: Wednesday morning by S. Resconi Photon: Thursday afternoon by T. Koffas 42
First Electrons Identified in ATLAS 2-tracks 1-tracks Electrons produced by ionization (δ-rays) Look for EM cluster with E>3 GeV loosely associated to a track Signal sample: events with 2 tracks Background: events with 1 track (muon Bremsstrahlung) Select on TRT signal and E/p TRT ratio>0.8 both 1&2-tracks 43
Event display of an electron candidate 44
Combined Muons φ difference θ difference d 0 difference Good agreement between track parameters measured in ID and MS Difference in momentum agrees with expected energy loss in the calorimeter (~3 GeV) p difference 45
Tau Identification Do not expect real taus in the cosmic data but it is nevertheless useful to check the tau identification variables reconstructed with real data isolation fraction : ratio of E T in a cone of 0.1< R<0.2 over E T in a cone of R=0.4) visible mass: invariant mass of the charged and neutral decay products, where the energy deposits associated to tracks are replaced by measurements in the Inner Detector 46
Jets Studies Jet transverse energy distribution for events triggered by the L1Calo trigger, compared to Monte Carlo EM fraction of jets with E T >20 GeV compared to cosmic MC and QCD dijet MC 47
Expectation of the first 100 pb -1 Example channels J/ψ µµ ϒ µµ W µν Z µµ tt WbWb µν µν+x QCD jets p T >1 TeV Expected statistics in ATLAS after cuts s = 10 TeV, 100 pb -1 ~ 1 M ~ 50 K ~ 300 K ~ 30 K ~ 350 ~ 500 1 TeV gluino, squark ~ 5 Goal in 2010: Commission and calibrate the detector in situ SM measurements at s ~ 7-10 TeV: W, Z, tt, QCD jets Early discoveries? big (good) surprises? 48
Summary Hundreds millions of cosmic events and one major access later ATLAS is very well prepared to receive first collisions. The next challenge is the final commissioning and calibration with collision data and to maintain a high operational efficiency. The era of exciting new discoveries is beginning. 49
Thank you! 50