Il Joong Kil

1
Status Update ATLAS Semi-Conducter Tracker
(SCT) Module Production at Melbourne

• Il Joong Kil
• AusHEP
• November, 2004

http//epp.ph.unimelb.edu.au/epp/sctprod/sctprod.h
tml
2
Talk Outline

• Introduction SCT Overview
• Where we were in 2003
• QA Processes
• 2004 - Pre-qualification - Qualification
• Melbourne SCT Modules
• Issues
• SCT Production 2005
• SCTDB
• Conclusion

Melbourne SCT Production TeamGareth F.
MoorheadJared WintonSteve GregoryScott Moncrief
3
ATLAS Inner Detector

• ATLAS Inner Detector
• Pixel Detector
• 3 layers, 8 disks
• r 4.8 - 16 cm, 140 M channels
• SemiConductor Tracker (SCT)
• (Silicon strip detector)
• r 27 - 52 cm, 6.3 M channels
• Transition Radiation Tracker (TRT)
• (Straw tube detector)
• r 56 - 107 cm, 330 k channels

Tracking and Vertex reconstruction Axial magnetic
field 2T
4
The Semi-Conductor Tracker (SCT) Layout

• 4 Barrel cylinders, 2 x 9 End-cap disks
• 4088 Silicon modules (2112 Barrel and 1976
  Forward (four different kinds))
• 61 m2 silicon, 6.2 M channels
• 15,392 Wafers
• Resolution Barrel ?(R?) 16 ?m ?(z) 580 ?m
• Resolution Forward ?(R?) 16 ?m ?(R) 580 ?m
• (From Technical Design Report)

? 1.0
? 1.5
? 2.0

• 9 Wheels each in 2 Endcaps
• 1.4 lt ? lt 2.5

? 2.5
5
Outer Endcap Module Components
6
By the end of 2003

• Assembly
• Lack of real hybrids
• CiS wafers are more delicate than Hamamatsu
• Persistent problem with silicon-spine
  registration (Phase 1)
• Metrology
• After the installation of Mitutoyu QV302
  metrology system (AUS70,000)
• The new metrology system aided us in better
  identifying mechanical discrepancies between our
  modules and the specifications
• Bonding
• Problems with bonding consistency. Our bonding
  procedure creates gt100 noisy silicon strips per
  detector (upper specification limit 12)
• Bonding was considerably time consuming
• General
• We were not pre-qualified by the end of 2003

7
Coming into 2004

• New bonder arrived in January (AUS250,000)
• Clean-room extension and temperature control
  (Jan-Feb)
• Assembly jig reworked to resolve Z metrology
  problem

8
QA (Quality Assurance) Processes
9
QA Metrology

• We are currently using the software developed by
  Tom Atkinson who worked tireless and without
  complaint (program called Tri-Metro aka I
  cant believe its not SAS)

All modules from 57 and onwards are mechanically
in specification
10
QA Wire Bonding

• Process of the bonding may introduce defects to
  the modules
• New wire bonder (HK Bondjet 715M), has been
  installed in January to combat this
• MEL58 and MEL60 were used to get a set of
  parameters for MEL61 which gave good results. The
  parameters used are still under discussion.
• Changing the wire type is being considered for
  the improvement of the bonding conditions
• Wire type being used 25 ?m AlW alloy wire
• Others with CiS wafers use 17 ?m wire

11
QA IV Tests
discontinued for the production

• 2 IV tests under consideration
• Reception test on the wafers before the assembly
• IV curve test of the module from SCTDAQ
• Both are being tested for the voltage range of
  0350V ensuring the current does not exceed 5?A
• Both tests are done at a temperature of 20? with
  a tolerance of 1?
• Specification keeps changing

12
QA Electrical Test (DAQ)

• Full Characterisation Test (ABCD Test)
• SCTDAQ is used
• Temperature of the module throughout the test is
  to be stabilised at 10? permitting 3? tolerance
  at the maximum
• Long Term Electrical Test (LTT)
• No LTT has been run yet (currently under
  development)
• The sequence of full characterisation is run for
  the extensive period of time (about 24 hours)
• The temperature of the modules being tested
  should be monitored and the test should be
  terminated once the temperature becomes unstable

13
QA RC Plot of MEL65
14
Pre-Qualified Module Qualification Candidates

• Pre-Qualified Module MEL61
• Pre-Qualification approved during July/August
• All the mechanical specifications have been met
• Only partial electrical test is done at
  Melbourne, the rest along with the thermal
  cycling test is done at Geneva
• Qualification Candidates
• 5 consecutive qualifying modules are required for
  the qualification
• MEL63 Bonding parameter modified, too many
  defects
• MEL64 Front side bad, but the back side is
  really good The bonding parameter used
  in the backside is our current fixed bonding
  parameters
• MEL65 The first module that is in electrical
  specification
• MEL6667 A few amount of defects but within the
  specification
• MEL6869 Built and in mechanical spec, but not
  tested electrically yet

MEL6364 not good
15
Melbourne Modules Metrology
Within these lines in spec.
16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
Melbourne Metrology Defects Summary
20
Issues resolved

• All the mechanical problems are now resolved
• to name few, XY survey, Z survey, assembly phase
  1?2 discrepancies
• Although changing wire type to 17 ?m is suggested
  by other institutes, we will try to achieve the
  spec. with our current wire
• Electrical testing
• High voltage card (HV) did not function correctly
• Fixed at late October by Gareth
• All the electrical tests were done with a bias
  voltage of 100V until recently corrected to 150V
  as in the specification

21
Issues to be resolved

• For both assembly and metrology systems, we are
  still getting some difficulties with finding the
  correct fiducials
• Bonder settings need to be confirmed
• Thermal cycling to be ready fairly soon
• Electrical testing
• Still unsure of the official specification
• Front-end tuning is it acceptable?
• Would VDC patch card replace pseudo-optical
  cable?
• Long term electrical testing set-up

22
What we expect in 2005

• Qualified
• Expected by the end of 2004
• Start production
• 100 modules by April, 2005
• we would like to increase our production to 6
  modules/week

23
SCTDB SCT Database

• Central database for the SCT registrations and
  test results is managed within the framework of
  the Oracle 9i database server
• Not all part of the database is accessible only
  limited permissions

24
SCTDB Upload

• Database entry of each module is uploaded to the
  database (called SCTDB) via either web or java
  applications
• Web each information is entered manually ?
  tedious!
• Java much quicker and better way to upload but
  the data files are needed prior to any uploads
  except for the electrical testing where SCTDAQ
  generates correctly formatted results

25
Java Upload Example