1. Calibration of the Electromagnetic Calorimeter of the CMS detector
G. Franzoni
University of Minnesota
T. Tabarelli de Fatis
Università & INFN Milano Bicocca
Calibration definition and targets
Calibrations at start-up
In situ strategy for 2010
Calibration and stability monitoring

2. Preamble General concepts to provide context
Pointing to areas that will be elaborated in following talks
Addressing areas where actions are needed

3. ECAL layout High resolution PWO crystal ECAL Barrel: || < 1.48
36 Super Modules
61200 crystals (2 x 2 x 23 cm3) – 26X0
Avalanche photo diodes
Endcaps: 1.48 < || < 3.0
4 Dee’s
14648 crystals (3 x 3 x 22 cm3) – 25 X0
Vacuum photo triodes
Preshower 3X0 (Pb/Si) 1.65 < |η| < 2.6
Monitoring in LHC abort gap:
Laser light injected in all channels
LED light in endcaps

4. Definitions Ee/ = Fe/ G i ci Ai [ +EES ]
Calibration aims at the best estimate of the energy of e/’s
Energy deposited over multiple crystals:
Ee/ = Fe/ G i ci Ai [ +EES ]
Amplitude in ADC counts Ai
Intercalibration: uniform single channel response to a reference ci
Global scale calibration G
Particle-specific corrections (containment, clustering for e/’s) Fe/
Preshower included in the sum in endcaps
There’s inter-play across the different terms and a strategy to dis-entangle

5. Status at startup Precalibrations ci: Global energy scale G:
Barrel:
0.3% on 9 SM (electron beams)
% on 27 SM (cosmic rays)
ECAL Endcaps:
6.5% (crystal LY  VPT gain) combined w/ local uniformity of splash events
Still a chance to improve with
Preshower
2% (cosmic rays)
Global energy scale G:
Tied to test beam (also ES)
Corrections: Fe/
Algorithmic corrections based on MC; η, energy and cluster shape dependent
Need to be tested/tuned in situ since dependent on material budget

6. What if LHC start tomorrow
Zee width
Hγγ width EB EB EE
Performance acceptable for most physics in EB, nearly in EE
Target:
Target precision: 0.5% set by H benchmark channel
Approach a.s.a.p. in view of  resonances

7. Fast in situ intercalibration methodsP5
Tier0
CAF
AlCaRecoProducers
Calibration Algorithms
RecHits
RawData
RecHits
HLTFilters
Dedicated HLT filters for for fast intercalibrations:
-invariance of energy flow within an const-η ring
0/η->γγ mass constraint
calibrations with AlCaRaw (RecHits) to increase yield for calibration
Both methods provide intercalibration sets in a few days of data taking
No need to go into express stream
AlcaRaw production and CAF workflows tested at CSA08 and CRAFT09
Performances demonstration still outstanding in endcaps:
Worse S/N for 0/η; need ~1 week of data, precision to be assessed
Phi-invariance: never reproduced results of CSA06 (1-3%)

8. In situ strategy
Derive intercalibrations ci from phi-inv. and 0/η (Marat’s talk)
Fix absolute scale G and corrections (η, ET and cluster shape dependent) Fe/ with electrons from Ze+e- (Riccardo’s talk)
ES calibration (mip) and EE-ES inter-calibration (Ming’s talk)
Long-term also other channels: isolated electrons Weν
There’s sufficient redundancy of calibration sources to disentangle interplay between G/Fe/ and ci :
Validation and combination of calibration sets (tools and procedures in Riccardo’s talk)
Release new sets for reconstruction as long as precision improves. Further sets for monitoring.
Ee/ = Fe/ G i ci Ai

9. In situ strategy Ee/ = Fe/ G i ci Ai 0 calibration
None of the in situ methods fixes inter-ring scale
η inter-ring scale and correction functions for ci can be fixed using precalibrations
Inter-ring scale known to better than 0.3%
Ee/ = Fe/ G i ci Ai
0 calibration

10. Stability of the ECAL response: transparency
ECAL response will vary, depending on dose rate:
Crystals transparency drops and recoveries
2010 run: transparency change expected in innermost crystals of EE assuming luminosity will reach L = 1031 cm-2s-1
Simulation of transparency:
L = 2 x 1033cm-2s-1)
Scenario comparable to (ECAL TDR):
1031cm-2s-1
rel. Crystal response
8 E29 and 1E31
Transparency variation measured via response R/R0 to blue laser pulses injected in each channel in the LHC abort gap (Adi’s and David’s talks)
Correction to crystal energies proportional to: (R/R0 )α
with α=1.5 BCTP crystals, α=1 SIC crystals

11. Stability of the ECAL response: VPT gain
VPT gain varies (Sasha’s talk):
‘Classic VPT effect’ induced by LHC on/off changes in cathode current; mitigated by LED constant pulsing to limit current excursions: on average 1%
Optimal pulsing strategy yet to be defined
Long term ageing: irrelevant in 2010
Rel. VPT gain
Black: load=10kHz, <IC>~0.25nA; 46 days
h=2.1 and L=2.5*1033cm2s-1
Grey : load=20kHz, <IC>~1.0nA; 134 days
h=2.1 and L=1034cm2s-1
Rel. VPT gain
~25%
8 E29 and 1E31
Response to blue laser/LED and orange LED sensitive to VPT gain changes
Correction to crystal energies simply proportional to monitored change (α=1)

12. Calibration and stability
Due to the different values of α, in general one needs to correct separately for transparency and VPT gain:
Ee/ = Fe/ G i TiVici Ai
Correction for transparency change Ti
Correction for VPT gain change Vi
End-to-end test applying monitoring correction in RECO: yet to be completed
Procedures of validation of monitoring corrections (within start of prompt reco)
Capability of monitoring with orange LED yet to be proven data need to establish if LED can provide monitoring of VPT gain alone.
Strategy for 2010:
Activate monitoring procedure based on blue laser only
Being VPT ageing negligible and classic VPT effect ~1%, acceptable using blue laser monitoring to correct for VPT and transparency with the same value of α=1.5

13. Conclusions Definitions and procedures in place
ECAL calibration at startup: acceptable for most physics analysis
Areas needing attention:
Performance in EE
Stability in EE