US CMS Silicon Tracker Project

1
US CMS Silicon Tracker Project
2
210 m2 of Silicon Strips
Outer Barrel (TOB)
End Caps (TEC 12)
Inner Barrel Disks (TIB TID)
2,4 m
5.4 m
volume 24.4 m3 running temperature 10 0C
3
Silicon Strips
500 mm high resistivity
Blue double sided
320 mm thick low resistivity
Red single sided
Strip lengths 10 cm (innermost) to 20 cm
(outermost) Strip pitches 80mm (innermost) to
200mm (outermost)
4
Module Components
Pins
Front-End Hybrid Flex-ceramic laminate
w/integral Kapton cable
Kapton cable
Pitch Adapter
Kapton-bias circuit
Carbon Fiber Frame
Silicon Sensors
5
Technology

• Single sided p/n
• Industry standard
• Mass producible at low cost
• Surface radiation damage
• Increases strip capacitance (noise)
• p/n ok after inversion if adequately
  over-depleted
• High Breakdown Voltages
• Specific design and processing rules for guard
  strip geometries
• Al strip layer acts as a field plate to remove
  high field region from Si bulk to Oxide

6
Sensors
Pitch adapter
Frames
Hybrid
Hybrids
factories
Brussels
Brussels
CF carrier
Strasbourg
US in the tracker
CERN
Perugia
Wien
Louvain
RU
Sensor QAC
Karlsruhe
Strasbourg
Module assembly
Lyon
Bari
Perugia
Wien
UCSB
FNAL
Bonding testing
Aachen
Karlsruhe
Strasbourg
Zurich
Wien
UCSB
Bari
Firenze
Torino
Pisa
Padova
Integration into mechanics
ROD INTEGRATION
TIB
-
TID INTEGRATION
PETALS INTEGRATION
Aachen
Louvain
Karlsruhe
Lyon
Strasbourg
Pisa
FNAL
Brussels
UCSB
TOB
assembly
TIB
-
ID
assembly
TEC
assembly
TEC
assembly
Sub-assemblies
At CERN
Pisa
Aachen
Karlsruhe
.
--
gt Lyon
TK ASSEMBLY
At CERN
7
Covered in this talk

• Status of Production parts
• For Modules Sensors, Hybrids and Module Frames
• For Rods Rod Frames
• US Readiness
• US Group Evolution past year and upcoming year
• Status of all production equipment and manpower
• Cost Performance and Schedule

8
Components Overview

• Stockpiling Parts Now ? but with some caveats
• Sensors (500 mm thick)
• SGS Thomson (ST) ? Many problems. Production
  stopped
• Hamamatsu (HPK) ? Excellent quality. Deliveries
  behind expected
• Sensor Frames from Belgium/Pakistan on track
• Problems over the past few years appear to all be
  worked out for now
• Hybrids from Cicorel/Hybrid SA/CERN ? critical
  path
• Several design flaws and processing quality
  issues uncovered.
• Last one now being resolved
• Rod Frames from Helsinki/CERN ? recent mistake
  found on some
• Various residual problems have mostly been found
  and addressed.
• One found this week but not expected to cause any
  delays
• More may arise with experience..

9
Timeline of ST Sensor Issue

• August 03
• US uncovered a problem with ST sensors
• December 03 (3 day sensor workshop at CERN)
• Re-probing 1000 sensors in US and EU indicates
  quality has degraded from original
• US group conclusion degradation in time
  possible chemical deterioration.
• January 04
• Place orders with HPK for masks and prototypes
• February 04
• ST agrees to significant changes in QC and stable
  processing with the aim of being re-qualified at
  July 04 tracker week. Also agrees to cut order
  from 18,000 to 11,000. CMS places order with HPK
  for 7000 sensors
• May-July 04
• ST delivers 1000 qualification sensors. US builds
  177 modules. Sees time evolution in at least 2
  modules. Sensor groups see time evolution in 5
  of sensors probed.
• Tracker week - July 04
• probing groups together with ST uncover
  definitive evidence of corrosion resulting from
  large phosphorous content in surface oxide.
• ST is not qualified by CMS.

10
1. Common Mode Noise (CMN)
Situation as of early04

• CMN effect features
• Group of noisy strips with a turn-on voltage at
  which all 128 channels show high noise
• can appear after thermal cycles
• often accompanied by other types of degradation
  such as pinhole development, more CMN
• Often correlated with high current

Later a second chip develops a high noise
channel which causes common mode noise Channel
previously only had a slightly higher noise (0.3
ADC)
11
2. Vacuum effect ? single strips !
with vacuum
As of early04
without vacuum
Strip 420 421 (4µA 15µA). Switching probe chuck
vacuum on and off switches these strips on and
off. Effect is reproducible. No visible defect
seen.
12
3. Peculiar IV Curves
late 03 early 04
OB2 sensors
Its like no diode Ive ever seen Gromit -
Wallace
13
4. Long term instability
30211334388607
As of early04
14
5. Structure in leakage current
Good sensor
As of early04
15
6."Dots and Stains" development
initial
3 hrs, no hum. nothing more
As of July 04 tracker week
30 min, 40 RH Þ stains on guard first usually
1h30, 40 RH Þ new stains on bias (not always
visible on video, see later)
16
"Dots and Stains" origin
Investigation by Strasbourg and Karlsruhe (with
help of the Fraunhofer Institute Chemische
Technologie)
The ratio of elements in white areas of
stains indicates the existence of
Aluminum-oxide Corrosion !
As of July 04
17
Confirmation by ST
Aluminum corrosion
Both dots and stains are micro-corrosions of the
aluminum surface. The mechanism that drives this
phenomenon can be the following Humidity reacts
with Phosphorus (present in a 4 concentration
into the passivation oxide) and forms an acid
(probably H3PO4), that corrodes a superficial
layer of Aluminum.
Passivation (1 µm)
Aluminum (2 µm)
Triple oxide layer (1.5 µm)
As of July 04
18
7a. Long term sensor tests

• 233 sensors tested 72h (room temeperature,
  r.h.25-30)

Failure rate of qualification sensors in 72 h
period is 5
As of July 04
19
7b. Long term TOB module tests
Maxed out ADC Bits at this point
nA
After 7 hours, bias current started to
increase New high noise channels seen in
subsequent tests
sec

• Dark marks on bias ring occur near high noise
  channels

20
7c. Long term TEC module tests
22 modules tested 1 module with current increase
during LT test
As of July 04
21
The Current Situation

• Need 18,200 thick sensors installed in CMS
  tracker
• 20,000 total (10 spares) originally all ST
• Shifted orders to HPK
• Winter 04 7,000
• Summer 04 4,500
• Autumn 04 5,200
• TOTAL of 16,700 (1,500 short of installation
  requirement)
• Agree to accept 3,000 sensors from ST
• Installation of at least 1,500
• HPK Shipments
• Started on schedule in June 04
• Did not yet reach levels expected

22
Original Schedule
  Jan-04 Feb-04   Mar-04 Apr-04 May-04 Jun-04   Jul-04 Aug-04 Sep-04   Oct-04
                           
Mask Production                          
Sensor Pre-Production                          
Sensor Qualification         240                
Sensor Production           540 1000   1500 1500 1500   1500
 Cumulative Production           540 1540   3040 4540 6040   7540
Sensor Acceptance             500   1000 1500 1500   1500

• Initial plan showed 1st 7000 sensors delivered by
  November
• Current 4000 delivered.
• Met with Yamamoto (v.pres.HPK) Oct. 11 at CERN
  (JI on video
• 3 problems identified (One month lost while
  analyzing problems)
• Poly Silicon operator error caused over-etching
• Backside SiO2 too thin caused high leakage
  currents
• Scratches due to a problem with automated
  handling devices
• Recent batches have 75 yield allowing 1300/mo.
  rate
• If achieve 85-90 then will deliver 1500/mo.
• Agreed if order placed by Jan. HPK can deliver
  all 16,700 by Oct. 2005
• Also discussed option to extend quantity by ?
  3,000

23
Timeline of Hybrid Issues

• Summer 2003 US finds broken cable traces
• US reviews handling and studies alternative
  handling schemes
• CERN finds breaks are widespread
• Vendor says design is fatally flawed
• New design implemented after only 2 months delay
• Winter 2004 US finds strange failure mode in
  modules
• US traces the problem to the hybrid
• CERN responds instantaneously halts all hybrid
  production
• Find vias are not properly plated, with breaks
  occurring at unknown rate
• US Halts production of TOB and TEC modules except
  for ST qualification
• Many TEC TIB modules already done in EU (small
  radius HPK thin sensor modules)
• EU continues building
• Summer 2004 Vendor bought out.
• Management serious about solving this problem,
  with better resources.
• 4 variations of design processed
• Autumn 2004 QC Engineer at vendor - all trials
  are highly successful!
• Week October 15 2004 CMS qualifies substrate
• Week October 22 CMS to qualify fully loaded
  hybrids

24
1.Hybrid Cable Problem

• Flex cable fragility
• Problem was quickly solved
• Good US/CERN relationship
• CERN relationship with vendor

25
2.Good Vias and Bad Vias
Example of a good via
Cu
26
US Hybrids Delivery Schedule

• Oct. 04 200 TB hybrids
• old processing, known via prob. hesitant to use
  with HPK sensors
• Nov.04 270 TOB hybrids
• not the most recent design but all passed the
  testing, and fraction was test with all passing
  extreme thermal testing
• Jan.05 200 TOB 100 TEC hybrids
• Ramping up to Apr. 05
• Monthly rates 700 TOB, 500 TEC
• Half of the TEC will be sent back to Europe after
  they are wire-bonded and tested at FNAL/UCSB/MEX
• Each of the 3 North American hybrid processing
  centers has a minimum sustainable capacity of gt
  24/d ? gt 1600/mo
• We can lose a hybrid processing center at any
  time without loss of hybrid throughput

27
Module and Rod Transportation

• Winter-Spring 03 CERN reports that modules
  arriving from US have huge numbers of damaged
  wirebonds
• US proposes a successful solution (encapsulate
  joints)
• CERN confirms
• Winter-Spring 04 Rochester studies find
  flexible mother cable in rod can damage module
  wirebonds in transport
• CERN/US engineers study problem and design Al
  stabilizers.
• Autumn 04 US Discovered error in cross-bar
  placement on roughly 50 of rod frames (type-H).
• Helsinki developing the repair method.
• US will ship back 40 type H rod frames for
  repair
• Large numbers of rods will be stockpiled in
  advance of full production of modules

28
US CMS Tracker Group

• Brown University
• R. Hooper, G. Landsberg, C. Nguyen, H. Nguyen
• University of California, Riverside (UCR)
• P. Gartung, G. Hanson, G.Y. Jeng, G. Pasztor
• University of California, Santa Barbara (UCSB)
• A. Affolder, S. Burke, C. Campagnari, F.
  Garberson, D. Hale, J. Incandela,
• P. Kalavase, S. Kyre, J. Lamb, R. Taylor, D.
  White technicians
• University of Illinois, Chicago (UIC)
• E. Chabalina, C. Gerber, L. Nigra, T. Ten
• Fermilab (FNAL)
• M. Demarteau, A. Ronzhin, K. Sogut, L. Spiegel,
  S. Tkaczyk technicians
• University of Kansas (KU)
• P. Baringer, A. Bean, L. Christofek, D. Coppage
• Mexican Consortium
• Cinvestav H. Castilla, R. Perez, A. Sanchez
• Puebla E. Medel, H. Salazar
• San Luis Potosi A. Morelos
• University of Rochester (UR)
• R.Demina, R. Eusebi, E. Halkiadakis, A. Hocker,
  S. Korjenevski, P. Tipton

19 joined group this past year (includes 3 UCSB
technicians) - now adding a few more post-docs
students 9 left the group (includes KSU plus
several from UCSB)
29
Preparations

• Good parts in large quantities are coming in
• Deliveries will not be smooth
• Meeting the schedule will require
• Higher than expected peak production rates
• Extremely robust and stable production lines
• Well trained personnel
• Previous proven capacity in US is 15
  modules/day/site
• Further capacity expansions
• Almost no further fabrication equipment needed
  and no expansion in test equipment required
• UCSB and FNAL have already completed these
  changes
• Achieve by extending work day (split shifts)
  and/or adding support personnel to major
  production tasks
• Rates now possible
• FNAL 18/day sustainable and 21-24/d peak
• UCSB 21/d sustainable and 27-30/d peak

30
Assembly Plates

UCSB Plates Fabricated (parts made) Commissioned (ready to be used) plates used in module production so far
TOB R-phi 7 7 7
TOB Stereo 3 3 3
TEC R5 R-phi 2 2 2
TEC R5 Stereo 2 2 2
TEC R6 5 5 5
TEC R7 2 2 2
FNAL Plates
TOB R-phi 5 5 5
TOB Stereo 3 3 3
Total 29 29 29

• Total of 29 plates in the US (capacity of 3
  modules per plate)
• UCSB setup to do TEC.OR.TOB in any given day
• All have been exercised and are ready for use.

31
US Production Steps/Status
Task Capacity Manpower issues Software Issues? Hardware Issues
Hybrid Bonding Thermal Cycle 84/d Mexico not yet online No No
Module Assembly gt50/d None No No
Module Bonding gt50/d None No No
ARC Testing gt50/d None No No
LT Testing 200/wk UCR post-doc search No No
ARC LED gt50/d None No No
Module Reinforcing gt50/d None No No
Rod Assembly gt6/d None No No
Single rod test gt6/d UCSB post-doc search Yes Possibly
Multi-rod burn-in 32/wk UCSB post-doc search Yes Possibly
32
Module Mechanical Precision

• 97 modules meet the current stringent geometric
  specs
• Few failures are just outside the relative
  angular requirement
• US now applies 2nd order corrections
• No new modules outside specs
• Production quality excellent!
• Single Sensor Modules
• 0.20 Faulty strips
• Introduced faults lt 0.1 rate
• Two Sensor Modules
• 0.55 Faulty Strips
• Introduced faults lt 0.1 rate
• Will be much lower w/HPK

Dx(Frame-Sensor) (mm)
Dx(Sensor-Sensor) (mm)
Dq(Frame-Sensor) (mdeg)
Dq(Sensor-Sensor) (mdeg)
33
Misalignments and PT Resolution
MC Study of effect of misalignments on pt
resolution single m sample, pT100 GeV
Mean 0.0 RMS 3.5 mm Min -7 mm Max 7 mm
Recent US modules
34
Hybrid Thermal Cycler/ARCS Status

• Recently upgraded code
• PLL forcing
• Drifting pedestal check
• Added xml file auto-upload
• UCSB, FNAL and Mexico City test stands are
  commissioned and ready

We have all ARCS equipmentspares we need
35
DAQ Equipment Status

• 2 fully equipped Vienna boxes at UCSB and FNAL
• UCR Vienna box has enough DAQ equipment for 4
  slot stand
• TPO needed for 6 slots
• 2 single-rod stands
• Just received enough oMUX cards so re-cabling
  between rod types unnecessary
• 2 multi-rod thermal cyclers
• Both MUXs have been used to test 5 rods tested
  simultaneously
• Have enough equipment to fully commission system
• Only 5 MUX cards DAQ spares missing

• To instrument UCR Repair Center have all
  critical spare components required in the US we
  need
• 2 TSC in production?
• 3 TPO in production?
• 2 eMUX boards
• 7 oMUX boards
• 5 VUTRI - in production
• 10 PAACB half are built, half being assembled
  now
• 10 hybrid-to-utri adaptors in production

36
DAQ Equipment Status II

• With current TSC complement
• With 1 failure we lose either
• 1 Vienna box
• 1 single rod stand, or
• 1 rod thermal cycler

• With current TPOs
• With 1 failure we lose either
• 70 capacity of a Vienna box
• 1 single rod system
• 1 multi-rod system
• Cannot run more than 16 APVs in UCR stand
• Without the additional MUX, VUTRI, PAACB,
  hybrid-to-utri adapter boards
• Cant run UCR LT at full capacity which is
  crucial to ops of US Repair center

Component shortages and failures ? potential to
severely limit production testing capacity
37
Backup Equipment

• Spare sensor and hybrid tools were produced at
  UCSB for UCSB, FNAL and Brussels.
• Upgraded OGP computer OS and OGP software
• Automated routine occasionally missed fiducial
  marks. The new software fixes this problem.
• Set up back-up gantry computers with spare U600
  controllers and expansion cards already
  installed.
• We purchased backup components for every piece of
  production equipment or tooling that, if it were
  to fail, would cause a reduction in production
  rates.

38
US Module Types
39
First HPK Module Results From UCSB

• 6 R6 modules built using new HPK sensors
• All 6 modules are perfect
• Not a single flaw
• IV profile as expected
• Turn-on at low voltage
• Plateau bias current 600-700 nA

40
UCSB TEC Production

• Miscellaneous info
• Have built all required types successfully R5S,
  R5N, R6, R7
• 25 shipping boxes (20 modules each) built
• All carrier plates (100 per type) and all
  wirebond fixtures complete
• Capacity
• Could saturate UCSB production capacity with TEC
  modules
• Will depend on need and availability of parts as
  well as TOB production parts availability and
  schedule
• Another step higher in production capacity (by
  extending work day via overlapped shifts)
• Bonding and Testing capacity adequate
• LT testing capacity limit is 100 per week
• Eventually will be mostly TEC (TOB burn-in
  shifted to rods) or sampled

41
Outstanding problems/issues

• DB stability
• For our production rates, we must automate all DB
  queries.
• Need to standardize and maintain stable all data
  structures
• We rely on data to be accurate and complete from
  all preceding processing of components and
  structures.
• Successfully collaborating with our intl
  colleagues
• Old or un-installable components
• Prefer to remove them physically from our
  production sites and to have them properly marked
  in DB
• Rods
• We have recently achieved major milestones with
  rods but we are not out of the woods. See below

42
Rods

• Rod assembly understood
• Rod Testing
• Single rod testing is under control
• Multi-rod
• Had many problems with software and hardware
• Recently achieved major milestone at FNAL
• Can now run maximum at capacity (8 SS rods or 6
  DS rods i.e. up to 72 modules) for 3 days with
  thermal cycles!
• Remaining
• Get UCSB multi-rod test stand operational at same
  level as FNAL
• Had problems with some hardware- now fixed
• To finalize fault finding tests
• Finalize Database info and transfer methods
• Need experience with many rods to determine if
  there are issues with components.

43
MechanicsTracker Outer Barrel
Rods before/after modules installed
0.9 m
Final Cylinders at CERN
Full Prototype Wheel (for MSGCs)
44
Mechanics Tracker Inner Barrel
Support mechanics CF space frames and/or
Honeycomb structures
45
MechanicsTracker End Caps
Digital Optical Hybrid
R6
Interconnect Board
Analogue Optical Hybrid
R4
Frontend Hybrid
R2
46
Substructures in Test Beams

• May 2003 Beam Test (Bunched 25 ns beams of muons
  and pions)
• Systems of 6-10 TIB, TEC, TOB modules
• Detector performance as expected!
• May 2004 Beam Test
• Multiple rods, petals, and shells
• Larger system integration tests
• Tracking tests
• Position resolution, hit efficiency

Beam Direction
47
The TOB Cosmic rack in the test beamin June 2004

• Michael Eppard (CERN)
• on behalf of TOB CERN
• 23rd July 2004

Michael Eppard
Test General Meeting
23.07.2004
48
S/N Module 4 _at_ 300V (PEAK)
S/N gt 32
49
ORCA reconstruction of tracks
Ivan Reid
50
Cost Performance

• Delays have cost us
• Recently extended production to Jan. 2006
• Net increase of 600k in project
• Other US costs
• Paid for Masks (NRE) at Hamamatsu to be able to
  transfer sensor order from ST
• 290k
• Misc. equipment for higher/more robust production
• 100K
• Anticipated costs
• US Tech. to work at CERN on hybrids for 6 months
• 50k (?)
• Currently schedule has no contingency

51
SST Schedule
Completion Jan. 2006 An aggressive
schedule Will be revised Assumes 500
hybrids/wk Actual 400 hybrids/wk
52
Summary

• No longer have a manpower shortage
• In process of adding some personnel at UCSB
• Have studied all possible threats to production
  stability
• Purchased or manufactured spares
• Further increased capacity to 50 modules/d
• Requires manpower ( 4-5 FTE total)
• Systems
• All stages of production have been exercised and
  are near to final except rod testing
• Multi-rod stands rapidly converging

53
Conclusion

• Strips have been at war with poor components
• ST sensors have too many uncertainties
• Switched to HPK!
• Hybrids problem is solved large deliveries
  starting early 05
• US Role has been extremely important
• We are doing everything we can do