CMS detector performance
Jesús Puerta Pelayo (CIEMAT)
(on behalf of the CMS collaboration)
XL International Meeting on Fundamental Physics
Benasque (Huesca, Spain)
30/May/2012
Outline
CMS in a glimpse
CMS yesterday
Design guidelines & assembly
CMS today
Subdetectors description & performance
CMS tomorrow
The CMS upgrade program
in a glimpse
General purpose LHC experiment
Compact, hermetic, fully
solenoidal design.
All central tracking and calorimetry
inside a superconducting solenoid
(B=3.8T)
Powerful external muon
spectrometer
39 countries,
169 institutes,
3170 scientists and
engineers
(800 students)
Fully operational since LHC start
Almost 6 fb-1 collected at 7TeV
3.3 fb-1 at 8TeV in 2012
Many interesting physics results…
AN OUTSTANDING MACHINE!
CMS design guidelines
High performance muon spectrometer & trigger system
 Very good muon identification and momentum measurement
 Trigger efficiently and measure sign of TeV muons dp/p < 10%
High energy resolution electromagnetic calorimetry
 ~ 0.5% @ Et ~ 50 GeV
Powerful inner tracking systems
 Momentum resolution a factor 10 better than at LEP
Hermetic hadronic calorimeter
 Good missing ET resolution
Above all, a powerful magnet to help measuring particles momenta
Under construction (2000-2008)
Detector assembly took place between 2000 and 2008
The idea of assembling the detector in several big pieces on the surface in parallel to
infrastructure works in the cavern strongly influenced the CMS design, as it was built in
“slices” that could be lowered with a special gantry
From 2000 to 2006 preparation works in the cavern took place, in parallel to detector
assembly on the surface hall & labs.
Lowered, closed and ready for beam in September 2008
Several cosmic rays data taking campaigns with parts & full detector before the first
collisions were CRUCIAL in order to commission the detectors, DAQ and trigger systems,
calibration, alignment etc.
The CMS subdetectors
Momentum / charge of tracks and
secondary vertices are measured in central
tracker (Silicon layers).
Energy and positions of electrons and
photons measured in a high resolution
electromagnetic calorimeter. (~ 0.5% @ ET
~ 50 GeV)
Energy and position of hadrons and jets
measured mainly in hadronic calorimeters.
Muons identified and momentum
measured in external muon spectrometer
(+central tracker) dp/p<1% @ 100GeV and
<10%@1 TeV
Neutrinos identified and measured
computing missing transverse energy in
calorimeters (hermeticity = good missing
transverse energy resolution)
The inner tracker
First experiment with full Si tracker system. Largest silicon detector ever built.
Pixel detector for precise reconstruction of secondary vertices
Strip Tracker with excellent tracking efficiency and resolution
(Δp/p<1% @ 100GeV)
Full coverage range |η|<2.4
• Pixel detector: 3 barrel layers, 2 disks each side ~ 2m2, 66 M channels (radius 4.3 - 10.2 cm)
• Strip detector 10-12 barrel and endcap layers, 198m2, 9.3 M channels, (radius 20 - 116 cm)
Tracking based on multiple iterations using combinatorial track finding
Modules aligned up to O(10μm) accuracy
Tracker performance
Fraction of active detector:
Pixel: 97.3%
Strips: 97.8%
Hit resolution depends on sensor thickness and strip pitch
(the minimum value is reached for an angle corresponding
to optimal charge sharing)
Percentage of dead channels: 3.1%
Single hit efficiency: >99.9%
Strips: 15 μm to 45 μm
Pixels: 9 μm to 35 μm
Tracking performance
For low momenta tracks, momentum resolution
is driven by tracker performance.
Resolution on transverse momentum measured
using J/ψ mass line-shape.
Sensitive to:
• Knowledge of the tracker material
• Alignment
• B field
• Reconstruction algorithms
In general good agreement with MC (~5%, some
deviation in the transition region of barrel to end cap).
Primary vertex finding
efficiency above
99.5% for
>2 charged tracks.
For PV with more
than 10 tracks
with average
pT>1.2 GeV: close
to 20 μm
Particle ID in TK energy loss
• Hit cluster charge proportional to energy
deposit.
• Calculated dE/dx along trajectory
measurement using the silicon
strips hits.
• Particle identification possible up till 1
GeV.
K
p
d
Deuterons are
missing
in simulation
K
p
d
Electromagnetic Calorimeter
High granularity, extremely good resolution, low noise, good uniformity/intercalibration
Lead-tungstate (PbWO4) scintillating crystals
optically coupled to Avalanche Photo-Diodes
(Barrel) and Vacuum Photo-Triodes (Endcaps):
• High density (ρ=8.3 g/cm3)
• Short radiation length (X0=0.89 cm)
• Small Molière radius (RM=2.2 cm)
• Fast response (80% in 25 ns @ 425 nm)
Radiation-hard (need transparency monitoring)
Energy resolution: (~ 0.5% @ ET ~ 50 GeV)
Two lead radiators (2 and 1 radiation
lengths thick), each followed by a layer
of Si microstrip detectors, instrument
both Endcaps, acting as as preshower
Calorimeters.
ECAL performance
Fraction of alive channels:
• ECAL Barrel: 99.2%
• ECAL Endcap: 98.7%
• ECAL Preshower: 95.1%
2 complementary methods for crystal calibration:
• φ-symmetry calibration: exploiting the energy-flow
invariance around the beam axis in minimum bias
events (for crystals at the same pseudo-rapidity)
• π0/η calibration: mass constraint on di-photon events
(inter-calibration and monitoring of energy scale)
Systematic uncertainty is 0.6%(1.5%) for
Barrel(Endcaps) (for electrons)
Photon energy scale agrees with expectations:
1%(4%) level in Barrel(Endcaps)
ECAL performance
Radiation induces wavelength-dependent crystal
light-transmission loss (w/o changes in
scintillation) by few percent within a typical run,
recovering during the inter-fill periods.
Transparency correction needed
Overall effect of single channel
intercalibration and transparency
Corrections on the Z -> ee invariant
mass peak in the ECAL Barrel
Stable energy scale was achieved throughout
2011 run after applying laser
corrections
Z⟶ee
Energy resolution
ECAL provides very good energy resolution down to low energies
Performance in agreement with expectations
At high ET the scale in the barrel region is now set by the π0 calibration (correct to
1%); 3% shift in the endcap region
Good
knowledge
of material
budget
Photon & electron reconstruction
TRIGGER:
Electron and photon selection
starts online
• L1 e/γ trigger efficiency (nominal
threshold: 15 GeV) for electrons
from Z decays
Di-electron invariant mass
spectrum (2010 dataset)
Resonances clearly visible
Hadronic Calorimeters
Compact & hermetic calorimeter, good segmentation and coverage (|η|<5.2)
Sampling Calorimeter:
Scintillator (active) & Brass (passive)
Barrel: 2 half barrels,18 wedges of 20º in Phi each.
Endcap: 2 plugs, 18 endcap wedges each
17 active plastic scintillator tiles / layers.
Total 20916 tiles
1368 Megatiles
(sheets of scintillator)
Δη x Δφ = 0.087 x 0.087)
Few additional layers, the outer barrel (HO), tail catcher sitting
outside the coil, ensuring no energy leaks
out the back of the HB undetected.
2 forward calorimeters (HF) positioned at either end of CMS,
measuring high pseudorapidity range.
(Steel absorbers with quarz fibers)
Jet angular resolution ~ 20 (30) mrad
in φ (θ) at ET ≥ 100 GeV
HCAL performance
HCAL sub-detector very stable over 2010-11
period with > 99% of HCAL channels live
Gain corrections are based on LED monitoring.
Phi-symmetry corrections are calculated with
and without LED gain corrections.
With LED corrections applied, 2011 phi
calibration is close to 2010.
Correction factor after imposing phi symmetry
Isolated hadron response:
By using isolated tracks, an eta-dependent
energy correction to the data up to |η| ≤ 2.4
can be obtained
Dedicated trigger selecting events with
isolated tracks with PT>38GeV
before corrections
after corrections
Particle Flow
Alternative to «standard» energy
reconstruction method:
In CMS, charged particles get well
separated due to the extense tracking
volume and high magnetic field (3.8 T)
CMS has an excellent tracking resolution,
able to go to down to very low momenta
(~few hundred MeVs)
CMS has also an excellent
electromagnetic calorimeter with good
granularity in multi-jet events.
Only 10% of the energy corresponds to
neutral (stable) hadrons.
Factor of two improvement in energy
resolution with respect to measurements
using calorimeter information only
The muon system
•
•
•
•
Efficient muon identification
Precise muon pt measurement
(improves tracker based results
above 200-300 GeV/c)
Provide effective standalone
trigger capabilities
ENDCAP
BARREL
Detectors interleaved with the
magnet yoke steel layers
Cathode Strip Chambers (CSC)
used in endcaps
(tracking detector / self
triggering / performs bunch
crossing id. / able cope with
large rates magnetic filed
inhomogeneities)
4 stations in the endcaps
(0.9 < |η| < 2.4)
Drift Tubes (DT) used in
barrel (tracking detector /
Resistive Plate Chambers
(RPC) in barrel/endcaps
(excellent timing
resolution / mostly used
for triggering
self triggering / able to
perform bunch crossing
id.)
4 stations in the barrel
(|η| < 1.2)
Muon detector performance
Individual
performance of
both tracking
detectors well
inside
specifications
Hit efficiency > 99%, resolution between 200
and 400 microns
RPC: Good overall efficiency above 94% for
properly working station both in barrel and
endcaps
Regional muon trigger
Trigger primitive information (for DT and
CSC) or single hits (for RPC) are used to
reconstruct full muon tracks crossing the
spectrometer.
The BX identification is performed with
very high efficiency.
A coarse pt and position estimator is
propagated to further trigger levels
Transverse momentum measurement
is critical as it is used later on (Global
Trigger) to perform final accept/reject
decision
Muon reconstruction
Muon Identification based on “tight” quality cuts applied to muon reconstructed by the
Particle Flow (PF) algorithm similar to ones used for high pt analyses
Relative Combined Isolation (rel. comb. iso.) computed by measuring neutral hadron + charged hadron +
photon energy deposits in a cone around the μ and dividing by muon pt (rel. com. iso. cut applied in plot
< 0.15)
Efficiency estimated using Tag & Probe for PF muons passing tight selection criteria
Muon objects
The complete muon object is built using
tracker tracks (TK muon)+ muon system
tracks (Std Muon), in order to build a
Global muon
Muon detector performing remarkably
Excellent resolution in the resonances
Y(1,2,3S)
CMS status (end 2011)
5.72 fb-1 delivered by LHC and 5.2 fb-1
recorded by CMS.
Overall data taking efficiency ~91%.
Average fraction of operational channels per
subsystem >98.5%.
Uncertainty on the luminosity determination 4%.
CMS NOW! (2012 performance)
After the winter shutdown, LHC started
delivering new collisions at √s=8 TeV.
So far LHC performance is remarkable.
First stable beams delivered: 5/Apr/2012
β*=0.6m, Nb=47 bunches, 1.3 1011 p
Peak lumi already exceeded 2011’s
Expecting ∫Ldt ~16 fb-1 in 2012
Max day 2012 = 190 pb-1
(full 2010 = 43 pb-1)
Pileup (the price to pay…)
Real CMS event with 20 PVx
… but still affordable…
As instantaneous L increases, the number of
interactions in a single crossing gets larger.
In 2011 number of reconstructed vertices after the
August Technical Stop increased by factor 1.5
(β*=1.5 m)
Fills start with ~15 pile-up interactions
CMS can deal with this: high granularity and
relatively low occupancies
Vertex reconstruction still quite linear with
luminosity
Triggers able to cope with this challenging data
taking conditions
Offline algorithms substract activity not coming from
event primary vertex
CMS tomorrow: The Upgrade program
Why?
LHC turned out to be able to deliver much more luminosity than foreseen
-LHC operation plans go beyond 2030
-Finalize parts of the detector that were foreseen but not funded (forward muon)
-Maintainability: substitute non optimal parts (HO photodetectors)
-Maintain or if possible, improve detector performance (4th pixel layer, S/N calorimeter, longitudinal
segmentation at HCAL, timing information)
-Insure minimal degradation with time and with radiation (pixel, tracker replacement)
LHC upgrade scenarios
CMS (Phase 1)
(2010-2021)
CMS2 (Phase 2)
(2021-2030?)
The upgrade scenario
PHASE 1
400 fb-1
in 10-12
years….
4 TeV
Splices, SEE
Injectors
(linac4?),
collimation,
crab cav HL
Long Shutdowns
PHASE 2
3000 fb-1 in 10-12 years…. Lpeak=1035 cm-2s-1
Pile-up 200@50 ns (100@25ns)
Nominal LHC
1034 cm-2 s-1
25 ns
20 pile-up
7 TeV/beam
Phase I highlights (I)
Muon system
-DT θ Trigger Board replacement (innermost station)
-Relocation of SC out of the cavern (new ROS, new TSC, new
DTTF)
New external RPC endcap station
New external CSC ring
Improvement of innermost endcap muon station
New beam pipe & pixel detector
Higher efficiency (3/4)
Lower trigger rates
Recover trigger 2.4<n<2.1
Make it redundant?
Features of the new detector:
• 4 barrel layers (3) and 3 endcap
disks (2)
• Inner layer closer to IP
• New readout chip with
expanded buffers
• Less material budget
• CO2 two-phases cooling
• DC-DC powering
• New technology better radiation
tolerance?
32
Phase I highlights (II)
HCAL upgrades
* HPDs had to be operated at low gain
(decommissioned HO large noise in B field)
*SiPM will allow
-Longitudinal segmentation
-Timing measurement for rejecting
backgrounds
-MIP detection
Trigger & DAQ upgrades
* Increase bandwidth (basically all systems will move from electrical to optical
transmission)
* Employment of new technologies (uTCA instead of VME): aim to redirect to
~standarized electronics (large FPGA, multiple optical inputs)
* n-year replacement of computing equipment (PCs and networks), enlarge computers
farms
Conclusions
• CMS detector has operated extremely well since 2008:
• Millions of cosmic data have been collected in preparation
for LHC startup at the end of 2009.
• Since 2010 CMS has collected almost 5 fb-1 of integrated
luminosity in pp collisions at 7 TeV, and over 3.3 fb-1 in
2012 at 8 TeV.
• No major problems occurred during the two years of LHC pp & HI
operation.
• Detector performance is excellent, within expectations.
• Extensive upgrade program already in progress
• … and all this reflects in very precise and extraordinary tool for
physics measurement, as you’ll see next…
BACKUP
CMS L1 Trigger
L1 Muon trigger based on track segments
in the three muon systems (DT, CSC, RPC)
• Timing information used to match to the correct
bunch crossing
• HLT L2 refit of hits using full granularity and
track
reconstruction in regions of interest (defined in
L1),
momentum cut possible
• HLT L3 Hit reconstruction in tracker,
reconstruction
from pixel seeds in regions compatible with L2
interest and primary vertex
• Different muon trigger paths (single and double
muon).
• Single muon path pT at 3 GeV at startup.
Allowed measuring inclusive
beauty production in low momentum range.
16
Trigger requirements adapted to increasing
luminosity.
To keep the trigger rate within the devoted
bandwidth an option is to
devote a large fraction to non-prompt dimuon
production. This will
be the beneficial for B- and quarkonium physics.
Level 1 Muon Trigger System
RPC hits
DT hits
DT Local Trigger
RPC Pattern
Comparator (PAC)
CSC hits
segments
CSC Local
Trigger segments
DT Regional
Trigger
CSC Track
Finder
≤ 4 mu (barrel)
+
≤ 4 mu (endcaps)
≤ 4 muons
Calorimetric L1
Trigger
e / photon / jets / Et / Etmis
/ Ht / MIP / ISO
Global Trigger
prog. combined logic
L1 Trigger Accept Signal
Global Muon Trigger
≤ 4 muons
≤ 4 muons
b tagging
b-Hadrons are produced in jets
Displaced vertices due to long lifetime (1.5 ps) and
Lorentz boost
Clear signature
Identification of semi-leptonic decays
Low momentum (3 GeV) single muon trigger
threshold at startup
Probe inclusive beauty production at low
momentum
Inclusive reconstruction: Jet + Secondary vertex
identification
Exploit high precision of pixel tracker and long bhadron lifetimes
Efficient reconstruction of secondary vertex for ET
jet>20 GeV
Excellent for b-jet studies at larger momenta
Inclusive secondary vertex finder as a powerful
tool
for angular correlation studies, see talk on Tuesday
by Ch. Grab
b-Hadron exclusive identification
Full reconstruction of the b-hadrons fourmomenta
Competitive performance in Bd,c,s→ J/ψ X
(J/ψ→μμ).
b-Jet tagging works in a large transverse momentum
range.
• Uses excellent secondary vertex resolution from pixel
detector
• Between 50-60% tagging efficiency for pT~100 GeV
with 0.1%
background contamination