the cms detector – lecture 2 -...

1July 23, 2009 The CMS detector – Lecture 2: getting ready for Physics

The CMS detector – Lecture 2preparing for Physics

Martijn Mulders CERN, Geneva

International School of Physics Enrico Fermi, “Radiation and Particle Detectors” July 20-25, 2009, Varenna


Lecture OutlineLecture Outline

YESTERDAY

CMS: design and construction

• Physics goals

• Challenges at the LHC

• CMS design

• Building CMS

• First beam September 2008 !

TODAY

CMS: preparing for Physics

• Commissioning with Cosmics

– Detector Performances

– Fixes and upgrades

– The Magnetic Field

• CMS as a Cosmics Detector


CMS... ready for beam in September 2008


and CMS today: closed and ready again!CMS... ready for beam in September 2008


Between September 2008 and July 2009:

• Directly after the LHC incident of September 19, 2008:– CMS ramped up the magnet to 3.8T

– >Month-long sustained Cosmic Run At Four Tesla (CRAFT)

– Opened the detector for winter repairs:• Fixed faulty modules, correct some swapped cables etc

• Installed the pre-shower detector

• Today CMS is again closed and– ramping up the magnet

– to take another cosmic run Jul 23 – Sept 6 (CRAFT'09)


CMS integration and commissioning campaigns

-6-

2006 : Magnet Test and

Cosmic Challenge

(MTCC) (at surface)2007 – Apr 2008:

Detector integration –

commissioning runs in pit

April – Sep 2008 : Cosmic

RUns at Zero Tesla

(CRUZET 1-4)

Sept 2008 :

First beam events

beam halo muons

Mar-Jul 2009:

2-3 days integr. Runs

Jul-Sep 2009 (expected) :

6-week CRAFT'09

Oct-Nov 2008 :

Cosmic Run at Four Tesla

(CRAFT)

full detector, 3.8 T field


Why commissioning with Cosmic Rays?

• Before discovering SUSY, extra dimensions, Higgs...

• Understand Standard-Model signals (W, Z, top, J/psi)

• Understand charged hadrons, jets, electrons, photons,

beam spot etc – with first collisions

• Integrated detector commissioning with cosmic rays– Learn how to operate full detector, software and train people

– Study detector response to muons

• Learn to deal with known challenges (eg alignment)

• Discover + fix -unexpected- problems ! (eg B field)

improve situation now, save time later, reduce phase-space of 'unknowns' for tackling future surprises

– Can do a lot (but not everything) with cosmic muons

potentialphysicsbackground


Cosmics and beam runs•4 global cosmics runs with B=0T•300 Hz rate•350 M events

350M events

CRUZET1CRUZET2

CRUZET3

CRUZET4


Cosmics and beam runs•4 global cosmics runs with B=0T•300 Hz rate•350 M events

350M events

CRUZET1CRUZET2

CRUZET3

CRUZET4

September 2008: ~1 million beam halo muons

“beam splash” events: ~100k muonstraversing CMS horizontally


Cosmics and beam runs 2008•4 global cosmics runs with B=0T•300 Hz rate•350 M events

ECAL

HCALMUON

TRACKER

•1 long cosmics run with B=3.8T•4 weeks data taking (370M events)•19 days with B=3.8T (290M events)

•87% have a muon track in the chambers• 3% have a muon track with tracker hits • 30000 events have a track with pixel hits

Cosmics Run At Four TeslaCRAFT B=3.8T

290 M events

350M events

CRUZET1CRUZET2

CRUZET3

CRUZET4


Detector Performances (and repairs + upgrades)


Muon Barrel: Drift Tubes Efficiency

CRAFT

Drift Tubes layer efficiency. Lower values correspond to groups of temporarily disconnected channels.


The hit resolution is computed from the residuals between the DT hits and the track segments in the muon spectrometer.

Typical values σ ~ 200 – 260 µm

Good agreement with MC

CRAFT

Hit residuals (cm)-0.2 -0.1 0 0.1 0.2

Muon Barrel: Drift Tubes resolution


Winter upgrade: finalizing L1 DT Track Finder

In September 2008, only the 'phi' projection of the L1 DT track finder was fully implemented (important for pT determination!). For the 'eta' projection, only so-called “course” eta determination was available (important for seeding L2 muon track finding). This has now been fixed:

Comparing η of the L1 trigger with η measured by a stand-alone muon fitted by offline reconstructionComparing η of the L1 trigger with η measured by a stand-alone muon fitted by offline reconstruction


Y (cm)

X (cm)

Single gap working region

0 20 40

CRAFT

Muon Barrel: RPC efficiency

x


Y (cm)

X (cm)

Single gap working region

0 20 40

x

CRAFT

Swapped cables(fixed during shutdown)

Muon Barrel: RPC efficiency


ME−4ME−3ME−2ME−1

LHC Beam2 passes through CMS from negative to positive z

123

56

4CSC Sectors

BEAM

HALO

Muon EndCap: CSC occupancy


Muon EndCap: CSC efficiency

Endcap Muon CSC chambers performed well during cosmics run.

Average measured efficiencies for each station/ring:

efficiency to obtaina single hit in a layer

efficiency to obtain a tracksegment in a chamber

CRAFT

Module number


Tracker: operational fraction

CRAFT

Tracker Status at the end of CRAFT:

– Strip Tracker• TOB(1) : 98%• TIB/TID(2) : 96.6%• TEC+ : 99.2%• TEC-(3) : 97.8 %

– Pixels• Barrel pixels : 99.1%• Forward pixels(4) : 94.0%

(1) 0.6% recovered after CRAFT(2) 1% at least recoverable(3) 1.7% recovered after CRAFT(4) 5% recovered during shutdown (power cables repair), 0.5% still recoverable


On track strip clusters Signal to Noise ratio corrected for track angle effect

Hit finding efficiency of Barrel and End-Caps layers (after masking of faulty modules)

CRAFT

Signal To Noise Ratio (Tracker Outer Barrel)

Signal To Noise Ratio around 30 for all detector parts.

Tracker module

Tracker: S/N and hit efficiency


CRAFT

•Distribution of Median of Residuals (cm)Sensitive to remaining displacement of module in the measured coordinate

Tracker Alignment May'09→


CRAFT

•Distribution of Median of Residuals (cm)Sensitive to remaining displacement of module in the measured coordinate

Tracker Alignment June'09→

JUNE: huge improvement by combining HIP and Millipede algorithms, running HIP on topof geometry produced by Millipede

23July 23, 2009 The CMS detector – Lecture 2: getting ready for Physics 23

Tag and Probe method• Tag : stand-alone upper muons pointing to the tracker near the origin (LHC-like tracks) • Probe : Reconstructed tracker track

TAG

PROBE

CRAFT

Cosmics Data

Cosmics MC

ε>99.5%

Tracker: track finding efficiency


Muon energy in the calorimeter

region of interest at LHC


p measured in the tracker dE measured in the ECAL lower halfdE/ρdx energy deposit matched to the track corrected for muon path length

Collision loss

Brem

sstr

ahlu

ng

Theoretical curve

Data

Tracker momentum matches well with ECAL energy loss, ECAL energy scale is correct

CRAFT

ECAL: muon stopping power


HCAL barrel energy: signal corrected for muon path length in HCAL

Event selection:Muon track matching in DT and Tracker20 GeV/c < Pµ < 1000 GeV/c Cosmic muons data: 200 K events MC: 15 K events

HCAL Test Beam 2006Pµ = 150 GeV/c

Mean signal = 2.8 GeV

CRAFT

HCAL: muon stopping power


Last CMS detector to be installed. Preliminary commissioning shows 99.9% functional.

Pre-shower installation:


Muon track resolution: tracker-only

CRAFT


Muon track resolution: high pT

Resolution for muons with pT>200 GeV/cDepends crucially on alignmentwith latest alignment, approaching the point wherethe combined fit starts to beat tracker-only:

CRAFT


CRAFT alignment compared to expectations

Reaching the level of alignment expected after 10/pb of data... before the first collisions!


EndCap Alignment with beam halo muons


Precise Mapping of the CMS Magnetic Field


The CMS Magnetic Field Map• An accurate Magnetic Field Map is very important

• Used in MC simulation, trigger, and offline track reconstruction

Solenoid

3 endcap disks 5 barrel wheels, 3 'layers'

Tail catcher

Beam axis

3.8T map used for CRAFT


The CMS Magnetic Field Map

tracker

Muon spectrometerMuon spectrometer

• Three “regions” of the field:

• CRAFT: First opportunity to probe B field directly in the iron of the yoke, using muon tracks !

– Tracker region: field is very well known (<0.1%) thanks to MTCC fieldmapper measurements and NMR probes

– From tracker-to-muon system: dominated by well-behaved field inside solenoid →CRAFT indicates agreement <0.5%

– In return yoke: complex field, hard to measure, hard to model ! Aim: <5% (barrel) and <10% (endcap), mainly for muon trigger


Field in tracker region

• Measured by FieldMapper robot at nominal 2.0, 3.0, 3.5, 3.8 and 4 T field during 2006 Magnet Test and Cosmics Challenge:

– ~60k measurements with 3D Hall probes in cylinder of r=1.73, l=7 m (covering most of the volume inside barrel hadronic calorimeter)

– Several scans with NMR probes

– Flux loop measurements in iron return yoke during magnet discharges


Field in tracker region• Finite element calculation of field in

entire detector volume performed (at 2, 3, 3.5, 3.8, 4 T)

• Field map provided for physics, which agrees with measurements in tracker volume < 0.1%

• Further analysis achieves analytical fit with “Maxwell's equations” which describes measurements to 0.005% in tracker volume (10x better than probe calibration!), by fitting probe gain, offset, positions + scale set by NMR

confirms quality of measurements

• NMR probes inside solenoid confirm agreement scale <0.1% between 2006 and 2008

3.8T map


CRAFT: probe B field in yoke with tracks

• Measure muon momentum and charge in tracker

• Compare to muon trajectory in muon system

• Properly take into account dE/dx (~10 GeV for muon crossing whole detector)

• Effects may be small compared to multiple scattering...


First Hint of a problem – during CRAFT

B field?

Structure at lowpT probably relatedto material effects not fully corrected forin this preliminary plot

Plotted is the meanvalue of Gaussian fitto R (vs tracker track pT):

Observed 'online' during CRAFT (Oct-Nov 2008)

In CRAFT data, stand-alone muon momentum appears to be over-estimated by 20%, compared to tracker track momentum:

୩୯July , ୨୩ ୨୦୦୯ The CMS detector – Lecture : getting ready for Physics ୨

Confirmation with muon segments

Significantly more bending in MC than in data…

µ+ µ-

Average is not exactly at 0, because of small mis-alignment betweenStation 3 and 4

alignmentcross-checkwith B=0

The difference between the delta-phi for mu+ and mu- is proportional to the integral of B*dl between the stations (and to first order not affected by alignment)

“Deficit” of B field in data:

Layer1 ~ 5-10%

Layer2 ~ 25%

Layer3 ~ 30%December 2008

Using muons with tracker-track pT=10 GeV


The CMS B field puzzle...

• Is it really true?

• Where does the B field go?

• How could we make this mistake? What are the consequences?

• Plan of action:

– Confirm effect with independent analyses

– Compare to old Hall probe and flux-loop measurements

– Check the finite-element model calculations

– FIX the magnetic field map

– Redo L1 muon trigger tables

– Install more Hall probes (?)


Independent, improved analysis

φ3propφ3

data

φ2propφ2

data

real track (data) propagated track (B from map)

MB2

MB3

Main cuts: tracks pointing to IP, pT 15-100 GeV

Analysis method:

3 23 3 2 2 3 2 3 2

3 23 2 3 2

[( ) ( )]( )

prop data prop data MB map trueT MB MB

prop prop MB mapT MB

p B Bp B

φ φ φ φφ φ

−− −

−−

− − − −=−

iron

• Instead of comparing to MC simulation, compare directly (for each muon) to the muon trajectory predicted by the muon propagator (using the B field map and expected dE/dx). Take muon pT and charge from the tracker track.

• Take the 'double' difference (mu+ - mu-)/2 to remove sensitivity to residual mis-alignment, like in previous method

• Measure relative difference in B scale between B_map and B_true in each iron layer, for each wheel, for each sector ==> CONFIRMS DEFICIT OF B FIELD up to 30% in outer layer barrel yoke


TOSCA finite element model: problem found

(~1 Wb/line)

TOSCA model describes x>0 half of CMSPrecise description of solenoid and magnetic materialsOnly “tunable” parameter is the magnet currentBut: changing the outer boundary of the model has an effect!

OLD (CRAFT'08)

NEW (MARCH'09)


Field Lines

Old Tosca Model (default during CRAFT'08)


Field Lines

New Tosca Model (March 2009)


Integrated Flux

Flux returned by stray field outsideCMS (small B field but large area)

Old TOSCA model

New TOSCAModel (march'09)

New models predict ~factor 2 more flux 'escaping' from iron return yoke


New TOSCA agrees better with data

• New TOSCA: better agreement with data. Overall scale discrepancy now:

~0% in layer 1 (was 10%)



• But: CRAFT results are precise enough to see deviations from phi-uniformity... ! ==> change Field Map design to incorporate this

• Deviations are understood and corrected for in latest CMS Field Map:


12-fold phi symmetry in CMS ?

Cryogenics chimney

Chimney for electrical cabling

feet

To good approximation the CMS iron return yoke is phi-symmetric (12 equivalent sectors), apart from two chimneys and feet


Chimneys...


New default CMS Field Map

• The New B Field Map for simulation, trigger and track reconstruction is based on the New TOSCA model (March 2009), with

– Special treatment of chimneys and feet of the iron yoke

– Additional phi-uniform scaling factors, derived from CRAFT data:

|B|

|B|

|B|

locationof the“chimneys”

differentfield inbottom 3sectors:

W±2 W±1 W0 L1 0.994 1.004 1.005 L2 0.956 0.958 0.953 L3 0.918 0.924 0.906

scaling factors:

~0%~5%~9%


Total improvement from Old to New Map:

Bdata/Bmap


Closure Tests

All sectors/layer/wheels 1.000 ± 0.001

All top sectors 0.997 ± 0.002All bottom sectors 1.002 ± 0.002

All inner iron layers 0.999 ± 0.003All middle iron layers 1.001 ± 0.002All outer iron layers 0.999 ± 0.002

All in wheel 0 1.001 ± 0.002All in wheel +/-1 0.999 ± 0.002All in wheel +/-2 1.00 ± 0.02

0.993 ± 0.004

0.995 ± 0.005 → incoming muons 0.994 ± 0.006 → outgoing muons

0.998 ± 0.0070.992 ± 0.0050.991 ± 0.006

0.993 ± 0.0050.994 ± 0.0050.98 ± 0.02

CRAFT Datawrt New B map(phi-uniform scaling factorsapplied)

CRAFT MC(no scaling factors applied)

'trivial' closure test → after applying scaling factorsre-doing analysis expected to give factors = 1

This non-trivial test confirms that method gives the correct B scale in MC: the bias in the method (if any) isless than 1% !

(0.5 ± 0.3%)/2

Non-trivial !For top and bottom to agree, effect on muon momentum of dE/dx crossing half the detector needs to be controlled < half the difference:


CMS as a Cosmic Muon Detector


CMS as a cosmic muon detector

• CMS was not designed as a cosmic detector

• But it has a very powerful muon system

– Count muons with large multiplicity

– Measure high momentum accurately

– Good pointing accuracy

• (Partly) shielded with 100 m of 'earth'


Cosmics Charge Ratio Analysis

First CMS physics measurement, based on MTCC data:

working on improved result with CRAFT data


Cosmics Charge Ratio Analysis• Goal: fill the gap between L3

Cosmics and Minos →• Simple analysis in principle...

• Requires very good understanding of

– charge identification

– trigger effects

– Alignment!

• Work in progressNote: CMSnot charge symmetric!


Cosmic Muon Reconstruction Performance

• Cosmic muon crosses the whole detector by → combining both “halves” of the detector in a single fit, outstanding resolution is possible

• Below: plotting the difference between muon-system-only (SA 1-leg) and tracker-only measurement for cosmic muons. The pT on the horizontal axis is from the tracker track:


Conclusion and Outlook

• CMS, and the other LHC detectors, are beautiful scientific machines, unprecedented in size and complexity, built to meet unprecedented challenges of physics at the LHC

• CMS made good use of Cosmic Ray data and integrated tests, achieving some extremely promising results:

– Excellent efficiencies and resolutions of sub-detectors

– Alignment precision comparable to 10 pb-1 of LHC data

– Mapping central field, 10x better than thought possible

– Field map in iron yoke, improved 10x from <30% to <3%

• Will continue to analyze CRAFT'08 (and CRAFT'09) data

• Expect a set of CRAFT performance papers this summer/autumn

• We believe CMS is ready for data taking, but only real data will tell

• Looking forward to first collisions at the LHC !!


AcknowledgementsAcknowledgements

A lot of the material in these slides was borrowed from presentations by CMS colleagues. Thus I would like to thank Marcin Konecki, Nicola Amapane, Sara Bolognesi, Chris Tully, Jim Virdee, David Cockerill, Francesca Cavallari, Albert de Roeck for the material they contributed and for useful discussions


– Backup Slides –


Some interesting CRAFT events

2 muons with

pT = 200 GeV



>3 parallel muons

100 GeV muon with

shower in DT:



Run 66739

Event 9451445

the cms detector – lecture 2 -...

Documents