NIU Workshop R. Frey 1
Reconstruction Issues for Silicon/Tungsten ECal
R. Frey
U. Oregon
NIU Workshop, Nov 8, 2002
NIU Workshop R. Frey 2
Outline
• ECal Physics Goals• Current implementations
SD TESLA
• The hardware constraints Resolution requirements
• What simulation studies do the detector prototypers (we) want the simulators (us) to do -- discussion
NIU Workshop R. Frey 3
ECal Goals
• Photons in Jets Id. with high efficiency and measure with reasonable E resolution
• … in a very busy environment. Demand eff>95% with high purity
• Photon shower imaging vertexing (impact param. resolution 1 cm) º→ Separation from nearby photons, MIPs, h-shower fragments
• MIP tracking (h , muons) Id. Hadrons which shower in ECal
• Reconstruction of taus (eg →→º→--mip)• b/c reconstruction – include neutrals in MQ estimate• e’s and Bhabhas (Lum. spectrum) – easy (readout dynamic range)• Backgrounds immunity
Segmentation Timing
4
SD Si/W • 5x5 mm2 pixel 50M pixels• For each (6 inch) wafer:
1000 pixels (approx) One readout chip (ROC)
• Simple, scalable detector design: Minimum of fab. steps Use largest available wafers
Detector cost below $2/cm2
Electronics cost even less A reasonable (cheap?) cost
M. Breidenbach, D. Freytag, G. Haller, M. Huffer, J.J Russell
Stanford Linear Accelerator Center
R. Frey, D. Strom
U. Oregon
V. Radeka
Brookhaven National Lab
5
Readout chip connections
Use bump-bonding technique to mate ROC to
array of pads on wafer
6
Pad
Silicon wafer
PCB
Aluminium
Cooling tube Cooling tubeVFE chip
1.3 mm
1.0 mm
0.5 mm
Thermal contact
Gluing for electrical contact
AC coupling elements ?
power line command line signal out
CALICE design with electronics inside detector
NIU Workshop R. Frey 7
Si Timing
• Dynamic range: MIPs to Bhabhas About factor 2000 range per pixel Want to maintain resolution at both
ends of scale
• Timing: What do we need? NLC: 200 ns bunch trains – Do we
need to resolve cal. hits within a bunch?
Bhabhas: 15 Hz for >60 mrad at 1034
What about 2-photon/non-HEP background overlays?
Exotic new physics signatures
Can try to provide timing for each pixel
Ramp
Threshold
Ref
Mux
12 bit ADC
Logic
8.3 ms
200 ns
High Gain
Low Gain
Shaper
Is ≈10 ns resolution sufficient ?
NIU Workshop R. Frey 8
What are the constraints from the hardware?
• Dynamic range OK• Transverse segmentation almost independent of cost within
reasonable range (watch thermal load) Segmentation < Moliere radius is OK
• Radiation damage probably non-issue• Timing perhaps possible with resolution of 10-20 ns• Moliere radius (9mm x 2)• Energy resolution ↔ long. sampling ↔ Money
More coarse with ECal depth Also: pattern recognition implications
NIU Workshop R. Frey 9
e+e-→jj, 200 GeV; LCDRoot FastMC
• Perfect pattern recog.
• 0.01/sqrt(E) 0.01 (EM)
• 0.01/sqrt(E) 0.01 (HAD)
← 0.10/sqrt(Ej)
← 0.11/sqrt(Mjj)
NIU Workshop R. Frey 10
EM: 0.12/sqrt(E) 0.01HAD: 0.50/sqrt(E) 0.02
0.18/sqrt(Ej)
0.19/sqrt(Ej)
EM: 0.20/sqrt(E) 0.01 HAD: 0.70/sqrt(E) 0.02
0.20/sqrt(Ej)
11
E > 0.5 GeV
0.19/sqrt(Ej)
E
Eh0
E > 1 GeV, Eh0 >1 GeV0.20/sqrt(Ej)
E > 2 GeV0.20/sqrt(Ej)
NIU Workshop R. Frey 12
What simulations studies do we need?
• EFA tuning ↔ segmentation -MIP separation , tau, pi-zero reconstruction
• Background overlays ↔ timing requirement• Longitudinal sampling
EGS4 Geant4
• Distribution of hit occupancy in a detector wafer