Tracker Material and EGamma

1
Tracker Material and EGamma

• Yuri Gershtein

2
Introduction

• Every experiment that tried to do tracking and
  calorimetry at the same time had to deal with
  tracker material
• every experiment tries to get all the material
  into the MC and all underestimate it at first by
  50 to 100
• Effects caused by material
• loss of efficiency and resolution
• energy scales depend on ID cuts
• electron v.s. photon energy scale
• At CMS
• very large amount of material
• very strong magnetic field (curler pT 0.8 GeV)
• very few measurements per track (hard to do
  pattern recognition)
• Small compared to resolution
• given stand-alone ECAL resolution many effects
  negligible in other experiments are not small

3
Simulation

• First test uniformly increase density of all
  tracking components by 50
• OSCAR_3_6_5
• edit materials.xml
• generate single electrons and photons at 40 GeV
  with and without extra material
• ORCA_8_7_3
• barrel, with cracks cut out

4
40 GeV Photons
50 extra material
Default material
ET (supercluster)
ET (supercluster)
eta
eta
ET (supercluster)
ET (supercluster)
5
40 GeV Photons

• Effect is consistent with 15 loss in
  efficiency
• this is the most optimistic number the tighter
  the cuts, the larger would be the difference

Histogram default material Points extra
50
E3x3 / ET(supercluster)
ET(supercluster)
6
40 GeV Electrons

• Electrons radiate from the start, so the effect
  on them is much larger

Default material
50 extra material
ET (supercluster)
ET (supercluster)
eta
eta
ET (supercluster)
ET (supercluster)
7
40 GeV Electrons
Histogram default material Points extra
50
Energy resolution for narrow clusters is still
significantly worse
E3x3 / ET(supercluster)

• Drop in efficiency is especially large at higher
  rapidity - will influence the amount of data
  needed for W-e calibration

ET(supercluster)
8
Material Measurement

• Not only the total material spatial
  distribution too
• material close to IP is more dangerous (magnetic
  field!)
• Total material can be measured in several ways
• Mass v.s. pT for resonances decaying into MIPs
  (like J/????)
• pT(end)/pT(begin) of electrons fitted with GSF
• Spatial distribution
• counting converted photons
• Main problem
• Know efficiency of conversion reconstruction v.s.
  R it is not very likeley to be described by the
  MC
• In other experiments (CDF, DØ) outer tracker is
  lighter, so electrons can be tracked relatively
  easily, but in CMS tracker gets heavier with
  radius

9
Conversion reconstruction
But the number of reconstructed conversions
strongly depends on the amount of material here,
too
Want to sample material here

• Very tough to get rid of biases
• One method
• Try to split the task in two probability for an
  electron not to brem too much and probability
  to reconstruct the track of such electron
• For the first, use Z-ee with one electron
  identified and well-measured and use the second
  one as probe start tracking from outside in
  and measure average pT(R)/pT(ECAL)
• For the second use ?0?????ee

10
Conversion Reconstruction

• ?0?????ee
• three isolated EM clusters close in rapidity
• Azimuth positions of the clusters from electrons
  and their energy are enough to predict conversion
  point
• Combine electron and positron and construct
  invariant mass of ?(ee)
• look at the invariant mass distribution for all
  events, events which have one of electrons with a
  track, both, and when conversion vertex is
  reconstructed
• From ratios of those yields one can extract
  tracking and vertexing efficiencies.

11
MC particle Level

• Single ?0 sample 5 GeV (also looked at 10 and 20
  GeV)
• Trace particles to the calorimeter
• Require one of the ?0 photons to reach
  calorimeter unconverted
• Require both electrons from conversion of the
  second photon to reach the calorimeter
• Sum up pTs of all other photons/electrons (they
  are radiated from the original conversion
  electrons)

?0 mass
E_loss/E_ee
E_loss / E_ee
?0 mass
12
Now to the ECAL response

• Most untrivial decide which of the two clusters
  are from the electrons
• Intuitively those with extreme values of ?
• Doesnt work well at low pT choose the pair
  closest in ?
• Some more optimization needed

13
Pions in QCD events

• DC04 QCD bin 0-15 (only 2000 events)
• No MC truth used
• Shape cuts need tuning

Effective mass (gee)
14
Summary and Next Steps

• Material is usually severely underestimated at
  start-up
• Will manifests itself by
• lower then expected efficiencies
• lower then expected energy response and
  resolution
• shower shape not agreeing with MC
• Next steps
• Make realistic estimates of how many ?0 can be
  reconstructed this way
• Study precision of the pT(begin)/pT(end) method
• other ideas?