l. greiner 1fee 2014 – star pxl vertex detector star hft lbnl leo greiner, eric anderssen, giacomo...
Embed Size (px)
L. Greiner 1FEE 2014 – STAR PXL Vertex Detector
STAR HFTSTAR HFT
LBNLLeo Greiner, Eric Anderssen,
Giacomo Contin, Thorsten Stezelberger, Joe Silber, Xiangming Sun, Michal Szelezniak, Chinh Vu,
Howard Wieman, Sam Woodmansee
UT at AustinJerry Hoffman, Jo Schambach
PICSEL group of IPHC-Strasbourg (Marc Winter et al.,)
Experience from construction and operation of the first Vertex Detector
based on MAPS
2FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTTalk Outline
• STAR vertex detector upgrades at RHIC.• Pixel detector design and characteristics.• Sensors and mechanics.• Detector assembly and integration with lessons
learned.• Installation for 2014 run and first results.• Lessons learned and outlook.
3FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTPXL in STAR Inner Detector UpgradesTPC – Time Projection Chamber(main tracking detector in STAR)
HFT – Heavy Flavor Tracker SSD – Silicon Strip Detector
r = 22 cm IST – Inner Silicon Tracker
r = 14 cm PXL – Pixel Detector
r = 2.8, 8 cm
We track inward from the TPC with graded resolution:
TPC SSD IST PXL~1mm ~300µm ~250µm <30µm
Direct topological reconstruction of Charm
4FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTPXL Detector Design
Mechanical support with kinematic mounts (insertion side)
Insertion from one side2 layers5 sectors / half (10 sectors total)4 ladders/sector
Aluminum conductor Ladder Flex Cable
Ladder with 10 MAPS sensors (approx. 2×2 cm each)
carbon fiber sector tubes (~ 200 µm thick)
5FEE 2014 – STAR PXL Vertex DetectorL. Greiner
2 m (42 AWG TP)11 m (24 AWG TP)
100 m (fiber optic)
Highly parallel system
4 ladders per sector 1 Mass Termination Board (MTB) per sector 1 RDO board per sector 10 RDO boards in the PXL system
RDO motherboard w/ Xilinx Virtex-6 FPGA
RDO PC with fiber link to RDO board
Mass Termination Board (signal buffering) + latch-up protected power
PXL Detector Basic Unit (RDO)
Clk, config, data, powerClk, config, data
PXL built events
Trigger, Slow control,Configuration,etc.
Existing STAR infrastructure
6FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTDetector Design Characteristics
DCA Pointing resolution (12* 24 GeV/pc) m
Layers Layer 1 at 2.8 cm radius
Layer 2 at 8 cm radius
Pixel size 20.7 m X 20.7 m
Hit resolution 3.7 m (6 m geometric)
Position stability 6 m rms (20 m envelope)
Radiation length first layer X/X0 = 0.39% (Al conductor cable)
Number of pixels 356 M
Integration time (affects pileup) 185.6 s
Radiation environment 20 to 90 kRad / year
2*1011 to 1012 1MeV n eq/cm2
Rapid detector replacement ~ 1 day
356 M pixels on ~0.16 m2 of Silicon
* Simple geometric component, cluster centriod fitting gives factor of ~1.7 better.
7FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTPXL detector Ultimate-2 Sensor• Reticle size (~ 4 cm²)
• Pixel pitch 20.7 μm • 928 x 960 array ~890 k pixels
• Power dissipation ~170 mW/cm² @ 3.3V (air cooling)• Short integration time 185.6 μs• In pixel CDS• Discriminators at the end of each column (each row
processed in parallel)• 2 LVDS data outputs @ 160 MHz• Zero suppression and run length encoding on rows
with up to 9 hits/row.• Ping-pong memory for frame readout (~1500 hits
deep)• 4 sub-arrays to help with process variation• JTAG configuration of many internal parameters.• Individual discriminator disable, etc.• Built in automated testing routines for sensor probe
testing and characterization.• High Res Si option – significantly increases S/N and
radiation tolerance.• Sensors thinned to 50 µm.
Developed by PICSEL group of IPHC-Strasbourg
Optimized for the STAR environment
8FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTPXL Ladder AssemblyPrecision vacuum chuck fixtures to position sensors by hand.
Sensors are positioned with butted edges. Acrylic adhesive prevents CTE difference based damage.
Weights taken at all assembly steps to track material and as QA.
Cable reference holes for assembly
Hybrid cable with carbon fiber stiffener plate on back in position to glue on sensors.
9FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTSector and detector half assembly
SectorsLadders are glued on carbon fiber sector tubes in 4 stepsPixel positions on sector are measured and related to tooling ballsAfter touch probe measurements, sectors are tested electrically for damage from metrology
Detector halfSectors are mounted in dovetail slots on detector halfMetrology is done to relate sector tooling balls to each other and to kinematic mounts Detector half mapped
Sector assembly fixture
A detector half
Sector in the metrology setup
10FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTYieldssensor yield
sensors after thinning probe testedafter probe testing (tier 1&2) Probe testing yield
batch 1 1152 1003 1003 464 0.46
batch 2 1152 1139 1056 546 0.52
batch 3 1248 1174 357 217 (new probe card) 0.608
yield 0.934 0.508
after assembly + bonding
after sector mounting after metrology
Tot 113 103 85 53 48
Tested 92 59 53 48Good 84 54 48 48
yield 0.91 0.92 0.91 1.00
Production is still ongoing.
11FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTPXL insertion
PXL has a unique mechanical design. The PXL detector is inserted along rails and locks into a kinematic mount on the insertion end of the detector. This allows for rapid (1 day) replacement with a characterized spare detector.
Yes – we push it in by hand
Kinematic mounts Insertion of PXL detector
12FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTDetector Installation
Pxl installation pics1.Clean room2.Installed and cabled
PXL assembled in the STAR clean room
PXL inserted into the STAR TPC inner field cage, cabled and operational.Total installation time = 2 days.
13FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTLessons learned during construction and installation
• Sensors– Build testing functionality into the sensor design from the start optimized for probe testing/module testing.
We developed the parametric testing requirements first and implemented them in the design with IPHC.
• Readout – Co-develop the RDO with the sensor. It is then guaranteed to work and you have time to find and fix
incompatibilities and quirks.
• Probe testing– Proper probe pin design for curved thinned sensors.
– Spend time on whole probe system yield (ours varied between 46% - 60%).
– Administrative control of sensor ID worked well for us.
• Engineering run– Do a proper full system test (if possible) in the correct environment with partial full detector and
– We discovered mechanical problems even in a fully solid modeled system (interferences, kinematic mount insertion, etc.).
• Assembly– Spend time optimizing full yield through all production steps. Original ladder and sector yields were much
lower until all the problems were worked out.
– Acrylic adhesive works well as a CTE stress decoupling method.
Things that worked well and things that didn’t
14FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTCosmic Ray Running
• PXL was installed with all sensors (400) working. < 2k bad pixels out of 356M.
• All pixel positions on each detector half were mapped in CMM prior to installation.
• PXL RDO and integration with STAR DAQ, trigger, etc. complete and working well.
• Automated threshold setting scripts applied (1600 thresholds need to be set). Noise rate is ~1-2 x 10-6 per sensor for most of the sensors. Last few sensors were tuned manually.
• All parameters are stable and preliminary alignment with cosmic ray tracks was made.
• Some misalignment in the kinematic mounts (~1mm).
• Sensor positions on detector halves appear to be as measured in the CMM before installation.
15FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTPXL preliminary half-to-half pointing residuals
Consistent with expectations for alignment and momentum of muons.
Preliminary Preliminary Preliminary
Preliminary Preliminary Preliminary
Preliminary alignment by Alex Schmah
16FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFT15 GeV/c and 200 GeV/c running - status
15 GeV start
200 GeV start
• The HFT (PXL + IST) was used only intermittently in the 15 GeV/c Au+Au run.
• The HFT (PXL + IST) has been operated in the 200 GeV/c Au+Au since the start.
• The SSD is still commissioning and is expected to join the data taking in some weeks.
• We have 700M events stored and are on track to take >1G events in this run. This should meet the physics requirements.
End of run
Projection of current data rate to end of run
17FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTPreliminary DCA Pointing resolution
Preliminary750 MeV Kaon
750 MeV Kaon
* Alignment still not complete
200 GeV/c Au-Au dataSimulation with Al Cable
TPC + IST + PXL
18FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTLessons learned from operation
• Perform full LU/rad environment testing of thinned production sensors/modules for all expected conditions.
– Sensors were being damaged in radiation field. We have lost ~15/400 sensors.– Damage apparently halted by limiting the energy available to a latch-up event (set LU
current thresholds to ~120mA over the ladder running current).– This is still under investigation. We will be performing additional LU testing in the near
future with thinned production sensors.
• Have backup solutions – e.g. Cu rather than Al conductor cables– Our aluminum conductor flex PCBs had problems during fabrication and, due to late
delivery, only 2 ladders with aluminum conductor were installed in the first run of the detector.
• Implement as much remote configuration of sensor/module and detector operating parameters as possible to allow for remediation of surprises.
– We implemented remote setting of ladder LU thresholds, voltage supplied to the ladders, voltage read, current read after first engineering run tests.
– This allowed us to diagnose and stop the damage described above.
19FEE 2014 – STAR PXL Vertex DetectorL. Greiner
• The PXL detector at STAR was installed for the 2014 Au-Au run.
• The detector system appears to operate as designed, integration with STAR infrastructure is complete.
• The DCA pointing resolution performance of the installed HFT detectors appears to be as expected.
• Sensor damage related to the radiation field was observed. The damage appears to be able to be halted by using operational methods. Based on our observations, we should hopefully be able to prevent damage to the next installed detector.
• We expect to be able to deploy the spare detector (with Al conductor cable on the inner ladders) for the next run and repair the damage to the existing detector with the spare ladders being fabricated.
• At this point, it appears that the installed PXL detector will be able to complete the physics goals for this run.
• The spare detector should be ready in ~1 month.
• MAPS appear to be working well as a technology for vertex detectors.
20FEE 2014 – STAR PXL Vertex DetectorL. Greiner
• Extra slides
21FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTPossible Mechanisms• Non-ionizing (Neutron) damage? – Dislocation of atoms in the matrix generally
results in permanent damage to silicon. This is consistent with what we observe. In discussions with Matt Durham who worked on the Phenix FVTX detector, he indicated that the beam tune into Phenix had some component of scraping against a magnet or beam pipe surface that caused spallation neutron based damage in their detector and that the neutron rate was nearly 3 orders of magnitude above what was calculated. In our case, we have no information about the neutron flux at low radii from the beamline.
• Ionizing radiation damage? – The primary source of ionizing radiation damage is expected to come from the transit of MIPs through the sensors. This is expected to show up in hits registered in sensor pixels. The occupancy of charged tracks is approximately what was projected in the simulations ( ~300 hits per frame on the inner ladders and ~100 hits per frame in the outer ladders). The likelihood of normal beam activities causing the damage observed is judged to be low. There have been, however, a significant number of non-standard events during this run. Particularly during the 15 GeV run period.
• LU related damage? – It is possible that LU events could cause damage in the silicon. This would need to be a phenomenon particular to thinned silicon and/or high resistivity epi. We did extensive testing of full thickness sensors at the BASE facility at the 88” cyclotron at LBNL where we exposed sensors to many thousands of LU events to measure the LU LET onset and cross-section.
22FEE 2014 – STAR PXL Vertex DetectorL. Greiner
Digital current on the inner ladders
Before operational optimizations
23FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTPXL increased current consumption
24FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTProposed Testing Plan
Proton IrradiationThe testing done for ionizing radiation dose was done with gamma rays. This is a generally accepted method of assessing the radiation tolerance of silicon designs. The proton irradiation more closely mimics the environment at the operating STAR experiment and adds displacement damage to the mix. We propose to expose an existing powered PXL ladder to proton beams at the 88” cyclotron at LBNL at various rates up to 300 kRad as well as testing the latch-up cross section due to proton irradiation (if possible, the primary proton LU mechanism of energy deposition from recoil only turns on at ~100 MeV).Latch-up TestingIt is possible that the damage observed is related to latch-up. The fact that the sensors are now thinned to 50um could give opportunity for other failure mechanisms such as micro-fracturing due to higher LU point temperature excursions. We propose to expose an existing PXL ladder and thinned sensor on testing cards to heavy ion beams at the 88” Cyclotron BASE facility for up to 5k latch-up events on a sensor. We will also measure the voltage discharge profile on the ladder to understand the profile of energy deposition in the silicon LU area. The production sensors have high resistivity epi and the initial LU test sensors were standard epi.Neutron IrradiationIt is possible that we are being exposed to a much higher neutron flux than has been anticipated. We propose to irradiate some sensors to doses of 1013, 1014 and 1015 1MeV Neq / cm2 . These sensors have already been sent to Sandia for irradiation with a batch of ATLAS sensors.
25FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTRun 15090063 - Mon Mar 31 21:16:0
Sector 10 L1 = 1.57 AL2 = 0.78 AL3 = 0.78 AL4 = 0.78 A
26FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTLadder assembly work flow chart
Probe tested sensors
Electrically tested low mass cables
Electrically tested driver boards
Dimensionally checked composite backer
Ladder wire bonding
Wire bond encapsulation
Full functionality test
• Ladder characterization
• Reworking and troubleshooting
• Quality assessment• Initial validation
1 day without problem
sFull functionality testbias optimizationThreshold scanNormal readout mode testAccidental hit rate scan
Quick testThreshold scan @ nominal bias settings
27FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTSector/half-detector work flow chart
Full functionality test
Half detector headassembly
½ HFT PXL
Half detector head metrology
Full Half detector assembly
28FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTLadder Design
Ladder cable concept
29FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTRadiation length in low mass area
Si 50um (0.0529%)
acrylic 50um (0.0148%)
Encapsulant + bond wires (0.070%)
Capacitors + solder (0.0035%)
Al 30um – both sides (0.0248%)kapton 50um (0.0148%)
acrylic 50um (0.0148%)
Carbon composite 125um (0.0293%)
fromolderestimate Carbon composite 250um (0.1017%)
Si adhesive 100 um (0.0469%)
Total = 0.388%NOTE: Does not include sector tube side walls
30FEE 2014 – STAR PXL Vertex DetectorL. Greiner
STAR HFTPXL insertion mechanics
Interaction point view of the PXL insertion rails and kinematic mount points
Carbon fiber rails
31FEE 2014 – STAR PXL Vertex DetectorL. Greiner
• Probe card with readout electronics – derived from individual sensor test card
• Analog and digital sensor readout• Full speed readout at 160 MHz• Full sensor characterization at full speed
– Test results used for initial settings in ladder testing and PXL detector configuration
• 2nd generation probe card for production testing– only digital readout pins loaded
Yield modeling makes probe testing critical to the goal of assembling functional 10 sensor ladders.
We test thinned and diced 50 µm thick sensors (curved). This is not easy.
Assembling sensors into ladders – Probe Testing