The ATLAS Pixel Detector

1

• The ATLAS Pixel Detector
• Claudia Gemme INFN and University of Genova
• on behalf of the ATLAS Pixel Collaboration

2
The ATLAS Pixel Detector

• It is the innermost part of the silicon vertex
  tracker of the ATLAS experiment.
• It consists of two parts
• 3 barrel layers
• 33 forward-backward disks
• 2.0 m2 of sensitive area with 0.8 ? 108 channels
• 50 ?m ? 400 ?m silicon pixels (50 ?m ? 300 ?m in
  the B-layer)

3
Insertable Layout

• Pixel detector layout and design have been
  modified over the last year to cope with delays
  in radiation-hard integrated circuit electronics.
• Pros
• Complete Pixel detector can be inserted or
  removed with remainder of Inner Detector in
  place i.e. as late as possible for initial
  installation.
• The insertable concept will also facilitate
  maintenance, repair and upgrades.
• Reduced number of modules (-17)gt less time to
  produce them.
• Transverse impact parameter resolution is
  similar to that expected by the previous layout.
• Cons
• Smaller external radius (14.2cm?12.2cm).
• Worse material distribution (especially for high
  h).
• 2 of tracks with less than 3 points.

4
Insertable Layout

• 3 layers with radii 12.25 cm, 9.85 cm and 5.05 cm
  in the barrel.
• Radiation doses at the radius of the middle layer
  in 10 years of LHC operation are expected to be
  1015 1-MeV neq/cm2 total dose 50 Mrad. This is
  the dose to which individual detector components
  are designed.
• B-layer has particle flux 5 times higher than the
  middle layer it should be replaced every few
  years.

Barrel Transverse view
Longitudinal view
5
Local supports

• Barrel staves and disk sectors are Local Supports
  that hold and cool pixel modules.
• Many prototypes of Local Supports produced and
  tested (dimensional stability, thermal and
  pressure cycles) .
• Production Readiness Review for Local Supports
  and Conceptual Design Review of rest of mechanics
  passed in July01.
• The cooling fluid (C3F8) will flow in thin Al
  tubes (0.2mm) both for staves and disks. The Al
  tube in the staves has been introduced after
  roptures of the Carbon tube under very high
  pressure conditions (8 Atm).

6
Modules

• Modules are the basic building elements of the
  detector (1456 in the barrel 288 in the
  end-caps).
• Each module has an active area of 16.4 mm x 60.8
  mm.
• The sensitive area is read out by 16 FE chips,
  each serving a 18 columns x 160 row pixel matrix.
  
• The 16 FE chips are controlled by a Module
  Controller Chip (MCC).
• A Flex-Hybrid circuit glued on the sensor
  backside provides the signal routing between the
  16 FE chips and the MCC. It also provides power
  routing for the FEs, MCC and sensor.

7
Electronics Overview

• Four rad-hard ICs needed. Radiation hardness up
  to about 50 Mrad.
• MCC decode data/cmd signals (from DORIC),
  generate control signal for FEs, collect data
  from FEs and accumulate in FIFOs, check event
  consistency, build module event and sends to DAQ
  (via VDC), handle errors and fault conditions.
• FE chip control 18x160 pixels. Amplify sensor
  signal, on-chip data buffering in EOC FIFOs until
  trigger signal arrives, send data on serial link
  to MCC.
• VDC drive data off detector.
• DORIC regenerate clock and data/cmd signals.

8
Electronics history

• Rad-soft FE and MCC successful and used for
  module testing, irradiated sensor testing and
  demonstrating that ATLAS specifications could be
  met.
• Rad-hard FE development began with Atmel/DMILL in
  mid-98. These FE-D1 chips had very poor yield.
• Second run with Atmel submitted in July 00 and
  included two versions of FE (FE-D2D and FE-D2S)
  MCC, VDC and DORIC prototypes. FE-D2D as FE-D1
  with bug fixed. FE-D2s modified design to remove
  circuit elements sensitive to technology problems
  but at cost of removing some functionality.
• Yield of FE-D2S good enough to proceed to build a
  few active modules and single-chip. Single-chip
  devices tested at CERN H8 test beam (Aug01) and
  then irradiated. Next week they will be again at
  the H8 test beam to study irradiation effects.
• Due to the unsolvable yield problems, design work
  stopped for FE and MCC in DMILL in Sept 00 and
  all efforts concentrated on 0.25m process (Deep
  Sub-Micron, DSM). DSM and DMILL design work on
  VDC/DORIC continuing.
• First 0.25m Engineering run (MCC,FE,VDC,DORIC)
  foreseen in few weeks.

9
FE chip 0.25m

• Dual process approach with TSMC/IBM.
• First digital test chip submitted to TSMC in
  Jan01. Limited functionality but worked roughly
  as expected.
• Analog test chip contains preliminary designs of
  analog blocks of 20 pixels and other critical
  analog items. Submitted to IBM and TSMC in
  Feb/Mar01.
• TSMC chip delivered in May. IBM version
  (identical) delivered in July.

• General features as expected (noise, timewalk,
  trim DACs,ToT) except for thresholdgt very large
  dispersion.
• 1st irradiation with 55 MeV protons at very high
  dose rates (LBL cyclotron)
• On-line results indicate little or no change in
  performance after 50 Mrad.

10
MCC-History
MCC-AMS - 1998
MCC-D2 - 2000

• St. Cells FIFOs area (mm2)
• MCC-AMS 18000 32 56
• MCC-D2 13500 32 100
• MCC-DSM 30000 128 25

MCC-DSM - 2001
11
Sensor

• Baseline design
• n pixel in n-bulk material
• to operate partially depleted
• double-sided processing.
• Moderated p-spray isolation.
• Bias grid to allow testing before module
  assembly.
• Oxygenated silicon to improve radiation
  resistance and increase allowable time to room
  temperature (for repair/upgrades).
• 3 sensor tiles per wafer.
• Various test and monitoring structures.
• PRR completed Feb 00.

12
Sensor preproduction results

• Preproduction started in Aug00.
• Two vendors
• Cis preproduction complete. Test results show
  good quality and excellent correlation between
  our QC measurements and those done by CIS.
• Tesla they have had yield problems, but now
  believes that are solved. More wafers in next
  months.

• V_breakdown (the highest voltage for which the
  normalised current leakage current is lt 2mA)
  required to be gt 150V.
• We found that none of the tiles ruled good by the
  manufacturers (and paid for) turned out to be bad.

13
Modules History

• 98-99 Bare Module - MCC in package
• 2000 Flex Hybrid 1.x Module with PCB card,
  MCC e Flex
• 2001 Flex Hybrid 2.x - Module on CC support,
  no PCB card

14
Electronic Module Tests

• Test a module to check for
• connections of power, control and I/O signals
• operation of the 16 FE chips
• MCC communication protocol and event building
  capabilities
• hybridization quality comparison between
  vendors (Alenia Marconi Systems In bumps, IZM
  PbSn bumps)
• module performance (threshold, noise, bad
  channels).
• Testing procedure
• digital tests
• analog tests
• source measurements
• test beam.

15
Analog Tests

• Characterize threshold and noise of each channel
  by injecting a voltage signal in each pixel cell.
  
• Threshold and noise at different steps in
  assembling
• Tuning of threshold to have good uniformity in
  each FE chip
• Threshold stability.

noise(e-)
Scan voltage(mV)
Threshold(e-)
16
Analogue Results Threshold Uniformity
Mean Threshold 3200e
Before tuning threshold dispersion 290 e
After tuning threshold dispersion 190 e
17
Analogue Results Threshold Stability

• Both for AMS and IZM modules, the threshold
  distribution tends to drift by a few hundred
  electrons between threshold scans.

18
Noise plot

• Overall module noise 145 e
• Larger noise for pixel in column0 (pixel area 50
  x 600 mm2).
• Ganged pixels two diodes connected to the same
  readout cell in the interchip region.

19
Bad channels plot

• In this module 21 bad channels (16 merged, 5
  inefficient)
• Merged channels (i.e. bad channels coupled with a
  very noisy pixel) 0.1 i.e. 2 pixels/FEchip
• Missing channels (i.e. good electronics pixels
  but no signal with source) 2 10-5

c0,r159
5
6
20
Conclusions

• The ATLAS Pixel Collaboration is approaching
  pre-production and production for many items of
  the detector.
• Detector layout has been modified to cope with
  delays of radiation-hard integrated circuit
  electronics.
• Next year
• Begins production of sensors, Local Supports and
  mechanical structures.
• DSM electronics evaluation of first prototypes,
  second iteration.
• Modules production of 50 dummy modules per each
  of bump bonding vendors. They will be used to
  test the assembly procedure and to evaluate
  thermal and mechanical stresses.
• Flex Hybrid third generation prototype design
  completed, compatible with DSM chips.

21
Material

• To minimize material
• 250 mm thick sensor
• Electronics thinned to 150 mm
• all supports in carbon composite material it is
  ultra stable and ultra light (4.4Kg)
• Asymmetric distribution of material B-layer
  services exit on one side.

22
Coverage

• Coverage of the insertable layout
• Probability to have less than 3 layer hits
• 1.5 hlt1.5 (barrel)
• 2.5 hlt2.5 (end-caps)
• Almost independent of particle energy (down to
  0.4 GeV)
• Inability to have first disk close enough to
  barrel.