Rgis Terrier PCC Collge de France

1
GLAST calorimeter performanceEnergy, position
directionBackground rejection
2
Outline

• Calibration procedure
• In-flight calibration using cosmic rays
• Energy measurement
• Low energy correction for losses in the
  tracker
• High energy correction for shower leakage
• Position measurement
• Longitudinal position
• Transverse position
• Background rejection

3
Energy measurement Astrophysical drivers

• Energy range from 20 MeV to 300 GeV (up to 1
  TeV)
• Spectral measurements
• Straight and broken power laws
• Cut-off energies (pulsars)
• High energy lines (WIMPS?)
• Thin calorimeter (8.4 X0) with 1.1 X0
  tracker
• Wide field of view
• A large fraction of events are well
  contained

4
Calibration procedure

• Calibration in two steps
• Beam tests of a group of 4 modules on the ground
  to validate simulations
• In flight calibration using cosmic rays nuclei
  H, He, C, N, O, Ne, Mg, Si, and Fe spanning
  nearly full dynamic range of CAL readout
• Calorimeter measures deposited energy

• Elements of calibration process
• Extract multiMIP events.
• Identify likely Galactic Cosmic Rays
• Fit tracks
• Accept events with clean track through log
• Identify charges.
• Identify charge-changing interactions.
• Identify mass-changing interactions.
• Fit dE/dx.
• Accumulate energy losses and light asymmetries.

5
Calibration - Nuclei identification

• CAL module tested on GSI beam
• 700 MeV/A Ni beam
• polystyrene target upstream
• Nuclear species are determined
• even in presence of a spectrum

6
Energy Measurement below 500 MeV

• GLAST Tracker is 1.1Xo thick
• large fraction of energy never reaches CAL

• Use the tracker as a sampling calorimeter
• Find hits in a cone around fitted track
• cone opening angle 5 ?MS
• Around 90 of the hits in this cone are due to
  the track (as opposed to electronic noise)

Hits in upper layers
Hits in lower layers
Energy in Calorimeter
7
Energy Measurement below 500 MeV
68 Containment slope - 0.44
20
Fitted Gaussian slope - 0.59
Ecal
10
Ecorr
5
68 Containment slope - 0.33
20
10
Fitted Gaussian slope - 0.51
5
100
400
50
200
Energy (MeV)
8
GLAST Energy losses

• Shower containment in the calorimeter limited by
• - Losses in the tracker
• - Leakage from the back, cracks and
  sides of CAL
• Strong non-linearity of response at low and high
  energies
• Necessity to correct for
• - Longitudinal leakage
• - Shower Profile fitting
• Last layer correction
• Lateral leakage

Less than 30 containment at very high energies
(gt300 GeV)
Tracker energy loss (diffusion dE/dX)
For high energies (several tens of GeV) at least
30 leakage
9
Shower leakage - Last layer correction

• Minimizing global width on MC data
• on a layer by layer basis
• Contribution from last layer only
• Energy deposited in the last layer is
  proprtionnal to the number of escaping particles

• Energy estimate given by
• Ecorr f(Esum,?) Elast Esum
• f depends on deposited energy and angle
• Restore linearity and provides good energy
  resolution
• Works as long as shower maximum is contained

Last layer
68 containment
gaussian
0 30 45 60 losses in the
tracker
10
Shower leakage - Mean shower profile fitting

• Minimize
• Longitudinal energy density profile model
• Parameters
• E0 incident energy free
• z0 shower starting point parameters
• ? fixed to their
• ? mean value at E0
• Mean profile fitting

• Restores linearity over the whole energy range
• Gives energy measurement even when shower maximum
  is not contained
• Good energy resolution up to very high energies
  (20 at 1 TeV normal incidence)

Fit
Last layer
Sum
11
Energy reconstruction - Performances

• 1999-2000 SLAC beam tests
• Prototype GLAST tower CAL, TKR
• at end station A
• CAL module made of 80 crystals
• 2.1 cm thick

• For 20 GeV normal incidence electrons
• 7 (raw energy sum)
• 5 (profile fitting)
• 4 (last layer)
• Cf Do Couto e Silva et al. 2001

12
Energy reconstruction performances
Depending on energy and angle, different
corrections have to be applied
Fitted energy resolution for very high
energies At angles larger than 50, less than 6
resolution at 1 TeV
13
Position et Direction - Motivations

• Motivations
• - Hadronic background rejection
• - Pointing improvement for high energies
• - CAL only direction determination
• Hodoscopic array of crystals provides 1 position
  per layer
• Longitudinal for 1 crystal
• With x barycenter of energy deposition along
  crystal axis
• The crystal with highest energy deposition yields
  the best estimate.
• Transverse energy weighted mean of crystal
  positions
• Pb Both methods are biased.

14
Longitudinal position

• SLAC 97 Beam test
• 3x3x19 cm crystals
• High performance readout
• In practice, limited by
• electronic noise
• systematics

15
Position Longitudinal bias

• For large incidence angles, the barycentre
  position in a crystal is different from the
  shower axis
• effect due to the mixing of longitudinal and
  transverse shower profiles

Increasing profile
Shower Maximum
Decreasing profile
16
Position Transversal bias

• Bias due to layer segmentation
• (size of a crystal larger than lateral extension
  of the shower)
• baryrec ? barytrue
• Correction using the usual S shape function
• depends on depth in the CAL
• incident energy
• For non-zero incidence angles, previous shift has
  to be added

3rd layer
5th layer
7th layer
17
Position measurement

• After correction, longitudinal dispersion is
  twice as good as transverse dispersion

• Barycenter dispersion as function of energy

Transverse
Longitudinal
Errors after correction
Note that this doesn t include systematics due
to non linearity of preamps, charge injection and
asymetry calibration precision
18
Background rejection

• Background rejection 1106 required
• constraint comes from extragalactic gamma
  ray background
• ACD efficiency is 0.9997
• CAL TKR need to reject remaining background

CR protons
CR electrons
EGRB
e
Upward moving particles shower
orientation High energy hadrons (over a few
GeV) CAL is 0.5 ?int 40 to 80 incident
protons interact and produce a shower - Lateral
and longitudinal spread - barycenter position -
difference between CAL and TKR directions 0.99
efficiency easily achievable
p