CMS ECAL 2006 Test Beams Effort

CMS ECAL 2006 Test Beams Effort

Caltech HEP Seminar

Christopher Rogan

California Institute of Technology

May 1, 2007

May 1, 2007 Christopher Rogan - Caltech HEP Seminar 2

CMS Detector

Crystal ECAL General purpose detector

p-p collision at CM energy of 14 TeV

Goals: Discover the Higgs, new physics beyond standard model, …


State of the Higgs: 2007

Electroweak fit (w/ quantum corrections) to mH : depends on mW, mTOP

Best-fit value (2007): mH = 76+34–23 GeV

using mTOP = 170.9 ± 1.8, mW = 80.396 ± .025 GeV

Direct search limit: mH > 114.4 GeV

95% CL upper limit: mH < 144 GeV

Low MH < 150 GeV


ECAL layout

barrelbarrelSuper ModuleSuper Module(1700 crystals)(1700 crystals)

endcapendcapsupercystalssupercystals(5x5 crystals)(5x5 crystals)

Pb/Si preshowerPb/Si preshower

barrel cystalsbarrel cystals

EndCap “Dee”EndCap “Dee”3662 crystals3662 crystals

Barrel: Barrel: ||| < 1.48| < 1.48

36 Super Modules36 Super Modules61200 crystals (61200 crystals (2x2x23cm2x2x23cm33))

EndCaps: EndCaps: 1.48 < |1.48 < || < 3.0| < 3.0

4 Dees4 Dees14648 crystals 14648 crystals (3x3x22cm(3x3x22cm33))

PWO: PbWO4


CMS ECAL Test Beams 2006

H4 ECAL Test Beam 10 SM calibrated (1 twice, 13600 xtals) Detailed studies of E, behaviour Irradiation studies Energy linearity studies

H2 ECAL+HCAL Test Beam 1 ECAL SM Two subdetector DAQ Wide beam calibration 0 data

H4

H2


CMS ECAL Test Beams 2006

A wide array of important studies were completed:Electron, 0 and cosmic muon inter-calibrationsEnergy linearity studies Crystal containment correctionsEnergy resolution studiesAmplitude reconstruction optimizationNoise studiesDAQ, Monte Carlo and software studiesOnline laser monitoringCrystal irradiation


Cluster Containment Corrections

683 703 723

684 704 724

685 705 725

ExampleExample: 3x3 matrix: 3x3 matrix

5x55x53x33x3

Containment effect decreases with the matrix sizeContainment effect decreases with the matrix size

3%

Hodoscope

Resolution: Uniform impactUniform impact containment corrections needed

Measurement in fixed size matrix of NxN crystals Measurement in fixed size matrix of NxN crystals position dependence position dependence of Eof ERECREC

e

1


Energy Resolution

• Energy resolution ≤ 0.5% at 120 GeV for any electron impact.• Same shower containment correction applied (for all E and all Xtals).

0.5%0.5%

Central impact “Uniform” impact


Caltech CMS @ ECAL test beams

Caltech leadership in two important test beam tasks:

Operation of the online laser monitoring system

Improving π0 inter-calibration technique using test beam data


ECAL Laser Monitoring Introduction

CMS is building a high resolution Crystal Calorimeter (ECAL) to be operated at LHC in a very harsh radiation environment.

PbWO4 Crystals change transparency under radiation

Correct using the observations of laser monitoring systemThe damage is significant (few % - up to ~5 % for CMS ECAL barrel radiation

levels) at high luminosity

The dynamics of the transparency change is fast (few hours) compared to the time scale needed for a calibration with physics events (weeks - month).

Resolution design goal: ~0.5%

Calibrating and maintaining the calibration of this device will be very challenging. Hadronic environment makes physics calibration more

challenging


Laser Monitoring System

Lasers at two different wavelengths:1 = 440 nm

2 = 796 nm


Laser Monitoring System

Laser light is injected into the crystals via fiber-optic cables

Avalanche photodiode response is measured (APD)

Light is also injected in reference PN diodes

Ratio of APD and PN responses is used to monitor crystal transparency changes


Irradiation Crystal Response

Monte Carlo with a ~12 hour LHC fill cycle


Irradiation Crystal Response


Laser Monitoring @ H4

Test Beam at CERN from June to November 2006

One ECAL supermodule in beam at time

15-250 GeV electrons

Intensity: Up to 50K events / 60s, Approx. 15 rad/hour

Online monitoring system was implemented to reconstruct laser runs and log values

Moveable stand

ECAL SM 22

Beam line


Online Laser Monitoring

For each laser run:APD and PN pulses reconstructed

APD, APD/PN and PN distributions for each channel (1700 per SM) are fit and used to extract mean values

Similar distributions are monitored in geometric groupings (half SM, light modules); used for potential corrections

Correlations between different values (APD - APD/PN - timing, Chi2, etc.)

10 ECAL supermodules examined

Over 1,600 laser runs processed


Online Laser Data Analysis

~15 min. to process each laser run

Plots of various distributions are available online immediately after

processing. APD/PN values (among other

things) logged in database for higher level analysis


Consecutive run monitoring

Comparison plots between consecutive runs for the APD/PN and APD values are used to monitor short term stability and inter-run changes

-.003

.001

0.0

Runs 13061->13064 SM16

00013061-00013064

For example, this plot shows the relative difference in the APD/PN values, for each channel, between two consecutive runs. Almost all channels are stable

to within .5 per mille between consecutive runs


Online Monitoring Stability

APD/PN

All channels, all modules :Stability 1.4 % from gauss fit to peak.

Overall stability good, even at this basic level without any further

corrections.

Get APD/PN ratios for each channel, each SM

Normalize average APD/PN to 1 for each SM

Fit gauss to normalized APD/PN for each channel

Sigma of these fits is the stability

APD/PNStability:

Raw stability


Offline Monitoring Stability

Mean before and after correction : 0.180 % 0.088 %

Peak before and after correction : ~0.170 % ~0.05 %

Small systematic change in reconstructed APD value related to Peak timing.

Correct APD/PN ratios with a simple linear function of peak timing

Example for one SM (22)


Raw Monitoring Stability at H2

APD/PN vs. Time, 100 Channels (1040 – 1140, center Module 3).

Hardware intervention around t=2150 h, stability reasonable.

Black : APD/PN, averaged over 100 channels.

Red : T/20+1

Anti-correlation between temperature and APD/PN – as expected.

APD/PN shows ~ -2%/C0 temperature dependences – as expected.

Temperature correction based on thermistors

Raw APD/PN stability at reasonable level


Laser Pulse Width Correction

Reconstructed APD/PN ratio sensitive to laser pulse width

For normalized APD/PN ratio, ~2%/ns

Long-term pulse width stability ~1-2 ns


Pulse Width Measurement

Linear fit of the APD/PN-width dependence for each channel of each SM

Normalize APD/PN by the fit value at width = 30 ns Distributions and crystal maps for the slope,

intercept, chi2, etc. of the linear fits for the normalized APD/PN values

error bars blown up by a factor of 10

normalization value

Example

Sigma / |Mean| = 6.9(1)%

A total of 6 SMs have been measured.

Pulse Width Non-Linearity has little channel to channel variation !

All slope for one SM


Example Irradiation Cycle

Normalized laser and electron responsesXtal 168

SM 22

For each electron response point an interpolated laser response value is calculated


Example Correlation Plot

Relative electron

response

Relative Laser Response

Xtal 168

SM 22


Example Corrected Resolution

Xtal 168

SM 22

120 GeV electrons, 3x3 crystal matrix


Continuing Irradiation Studies

Hodoscope hits - entire irradiation period

Beam events distributed throughout

crystal

Sufficient statistics to explore variations in

electron response within crystal

Xtal 168

SM 22



Hodoscope hits - entire irradiation period

Reconstruct electron data for 25

different bins

Generate R-plot for each bin

Xtal 168

SM 22


Continuing Irradiation StudiesC. Rogan

Xtal 168

SM 22



Still statistics limited in outer bins

Can potentially be used for precision offline corrections


Laser Monitoring Outlook

Measured the APD/PN stability for individual channels on a large scale

Demonstrated reasonable online APD/PN stability; could be used for online electron response corrections

Achieved offline APD/PN stability for majority of channels with simple corrections. Further corrections are currently being studied

Demonstrated the ability to maintain resolution during irradiation


π0 Calibration Concept

Level 1 trigger rate dominated by QCD: several π0‘s/event Useful π0γγ decays selected online from such events Main advantage: high π0 rate (nominal L1 rate is 100kHz !) “Design” calibration precision better than 0.5%

Achieving it would be crucial for the Hγγ detection Reporting on studies performed with about four million fully simulated QCD events. Results given for the scenario of L=2x1033cm-2s-1 and L1 rate of 10 kHz (LHC start-up).

Data after L1 Trigger Online Farm 0 Calibration

>10 kHz~1 kHz


π0 Selection

Based on local, crystal-level variables — suitable for online filter farm. Kinematics: PT () >1 GeV, PT (pair) > 3.5 GeV and η < 1.48 (barrel)

Photon shower-shape cuts: S9/S25 > 0.9 and S4/S9 > 0.9 defined with

2x2, 3x3, and 5x5 crystal matrices (S9 is chosen as photon energy) Additional isolation cut optimized to remove showers with significant

bremsstrahlung radiation: want to select mainly unconverted photons

Trigger Tower (5x5 Trigger Tower (5x5 crystals)crystals)


Selection Results

π0 rate of 0.9 kHzrate of 0.9 kHz or 1,250 or 1,250 ππ00/crystal/day with S/B /crystal/day with S/B ≈≈ 2.0 2.0 High-rapidity regions suffer both in rate and S/B (31)


A Calibration Algorithm (of many)

Simple iterative algorithm (L3/RFQ Calibration)Simple iterative algorithm (L3/RFQ Calibration)

(wi fraction of shower energy deposited in this crystal)

Both photon energy and direction reconstructed using crystal level information (same as during selection).

After each iteration pairs are re-selected with new constants (typically 10-15 iterations to converge).

Miscalibration is done before selecting events (4%). Calibration precision defined as R.M.S. of the product of the final and initial miscalibration constant. Use only pairs from ±2σ window around fitted π0 mass


Calibration Performance

Precision is then fitted to N is the number

a=27±1% and b=0.20±0.25% of π0/crystal2

2

bNa

CC +=

σ


Calibration Studies in Test Beams

π0 decays produced through: π-+Al π0+X (11/2006)

Three different π- beam energies: 9, 20, and 50 GeV

Consider only 9x8 crystal matrix: about 140 π0 decays/crystal


Reconstruction of π0


Selection of π0 using S1, S2 ADC


First Resonance Observed by CMS

Clear improvement over the uncalibrated peak (L3 algorithm). For a precise estimate of the calibration precision: use the 50 GeV electron test beam data.

π0 from upstream scintillators


50 GeV e- peaks with TBS1 9 GeV constants


Calibration Precision with 50 GeV Electrons

For each crystal, electron energy spectra were fitted to a Gaussian.Distributions of the obtained peak positions for 9x8 crystal matrix:

Precision: 1.0±0.1% with 0.9±0.1% expected. Calibrationwith ~5 GeV photon works well for higher-energy showers!


π0 Conclusions and Outlook

Proof-of-principle was achieved with full detector with full detector simulation: crystal-by-crystal intercalibration to 1% simulation: crystal-by-crystal intercalibration to 1% should be possible after a few days at L=2x10should be possible after a few days at L=2x103333cmcm-2-2ss-1-1

Other methods are much slower and tracker dependent.Other methods are much slower and tracker dependent. Optimistic outlook for achieving and maintaining aOptimistic outlook for achieving and maintaining a ~0.5% precision. Many months of work on understanding ~0.5% precision. Many months of work on understanding the ECAL performance and non-uniformity at lower the ECAL performance and non-uniformity at lower energies energies (work of ~15 physicists from 4 teams).(work of ~15 physicists from 4 teams). Test beam study demonstrated a 1% calibration precision Test beam study demonstrated a 1% calibration precision with ~5 GeV photons: successfully used to reconstruct with ~5 GeV photons: successfully used to reconstruct 50 GeV electrons. 50 GeV electrons. No noticeable systematics.No noticeable systematics. (Many thanks to the entire H2 test beam team). (Many thanks to the entire H2 test beam team). Currently a lot of work is being done on developing filter farm tools for collecting π0 in situ at the LHC. Calibration of the endcaps is also being considered.


Test Beam 2006 Summary

Two successful ECAL test beam efforts (H4, H2)

Recorded invaluable data for upcoming LHC startup while demonstrating viability of ECAL performance expectations

Caltech continues its leadership roles in hardware/software development of the 0 inter-calibration and laser monitoring

Credit is due to the hard work of entire ECAL community

Documents

CMS ECAL 2006 Test Beams Effort