BTeV Silicon Pixel Detector
Test beam results 1999-2000
Talk given by G. Chiodini – FermilabResearch Technique Seminar - Fermilab – July 21, 2000
July 21, 2000 G. Chiodini - Fermilab 2
Overview 1. BTeV experiment
– Intersection region at the Tevatron– Silicon pixel vertex detector– Silicon pixel detector
2. Test beam setup– SSD telescope and DAQ – Pixel planes tested
3. Test beam results– Pixel calibration– Charge collection– Charge-sharing– Spatial resolution– Magnetic field measurements– 4 Plane pixel telescope
4. Further studies– 0.25m radiation tolerant FPIX2– Pixel module bench test– Mechanics and cooling
5. Conclusions
July 21, 2000 G. Chiodini - Fermilab 3
BTeV experimentBTeV is an experiment to study Heavy QuarkPhysics in the C0 IR of the Tevatron:
• Mixing• CP violation• Rare Decays• …
• 1.6T Dipole magnet centered on the IR• Forward geometry (1.9<5) • Symmetric two arm spectrometer• Detached tracks in the 1stLevel trigger
July 21, 2000 G. Chiodini - Fermilab 4
BTeV experimentIntersection region at the Tevatron
• <2 interactions> per BCO• High density of tracks• Hadronic decays
Luminosity=2cms
• 41011 b-hadrons/107sec• B0, B+, Bs (bBc)• Forward region
BR<1.510-5L
Resolution B decay length in BTeVAng. Correl. bb production
July 21, 2000 G. Chiodini - Fermilab 5
BTeV experimentBTeV silicon pixel vertex detector
• Decay time resolution (CP studies) • good spatial resolution• high statistics data samples
• 1st L trigger: prim. vertex and detached tracks • high resolution points• detector close to the interaction region• low combinations and noise hits
• Pixel vertex detector • space points (x,y,z)• radiation hard (21014 particles cm-2y-1)• low occupancy and noise
Hybrid pixel detector
Readout chip
Sensor
Planar pixel vertex detector
bump
July 21, 2000 G. Chiodini - Fermilab 6
BTeV experimentBTeV silicon pixel vertex detector –
global layout
Pixel frame end view
Pixel frame lateral view
1.482m
0.595m
cooling pipes
July 21, 2000 G. Chiodini - Fermilab 7
BTeV experimentBTeV silicon pixel vertex detector –
half plane and module
beam
Module• 1Gbyte/s • Rad-Hard components(maybe optical device out rad. Area)
Half plane• 20% overlap of sensors in a single plane• 0.89% radiation length per plane
Pin diode optical receiver
VCSEL optical driver
Data serializer
Control, monitoring, and, timing
track
July 21, 2000 G. Chiodini - Fermilab 8
BTeV experimentSilicon pixel detector - hybrid detector
• Independent development and optimizations of readout chip and sensor• Require 5000 bump-bonding per cm2 to connect the pixel cells to the readout cells• Bump-bonding of flipped chip
• 2 acceptable bump metal (10-5 <bump failure10-4): Indium (In) and solder (SnPb) • Under Bump Metal (Cr, TiW, Cu, Au, …): adhesion layer, diffusion barrier and oxide prevention • Bonding process: Indium Metal bump both side, room T, pressure Solder Metal bump one side, high T, reflow
July 21, 2000 G. Chiodini - Fermilab 9
Harsh radiation environment p-type doping high Vdep detector must operate partially depleted after irradiation
BTeV experimentSilicon pixel detector – sensor
0V
Surface current
path Si-vacuum
Inter-pixel isolation
p-side multi-guard ring
2
2dqNV eff
dep Vdep=depletion voltage , d=detector thickness, Si dielectric constant, Neff=effective impurity concentration
21310 cmfluence
type inversion
n+/n/p technology
July 21, 2000 G. Chiodini - Fermilab 10
BTeV experimentSilicon pixel detector – sensor
According to the ROSE (RD48) Collaboration a factor of 2 in the depletion voltage can be gain using oxygen enriched silicon
Sensor radiation hardness can be improved by defects engineering
July 21, 2000 G. Chiodini - Fermilab 11
Three generations of readout chip each one with a specific goal:• FPIX0 to establish a viable front-end
• Hewlett Packard (HP) 0.8m CMOS process• 12cols x 64 rows• analog voltage output• simple column based digital logic
• FPIX1 to establish high speed digital logic• HP 0.5m CMOS process• 18cols x 160 rows• 2-bit flash ADC in each cells• Complete column based digital architecture
• FPIX2 Radiation tolerant design • 0.25m CMOS process (DSM) with Rad-Hard rules (guard rings and enclosed geometry transistors)• Redesigned analog FE feedback and leakage compensation• Simpler and faster digital section than FPIX1• preFPIX2T (2cols x 160rows of FPIX2)• FPIX2 submission in few months
BTeV experimentSilicon pixel detector – FPIXn readout chip
July 21, 2000 G. Chiodini - Fermilab 12
• 132 ns bunch crossing (BCO)• Column-based and data driven architecture
BTeV experimentSilicon pixel detector – FPIX readout chip
FPIX1 detailed block diagram
July 21, 2000 G. Chiodini - Fermilab 13
BTeV Silicon Pixel Detector
Test beam results 1999-2000
J.A. Appel, J.N. Butler, G. Cardoso, H. Cheung, G. Chiodini,
D.C. Christian, E.E. Gottschalk, B.K. Hall, J. Hoff, P. A. Kasper,
R. Kutschke, S.W.Kwan, A. Mekkaoui, R. Yarema,
and S. Zimmermann
Fermi National Accelerator Laboratory
C. Newsom - University of Iowa
A. Colautti, D. Menasce, and S. Sala - INFN(Milan)
R. Coluccia and M. Di Corato - Universita’ di Milano
M.Artuso and J.C. Wang – Syracuse University
Particular thanks toW. Baker, C. Brown, J. Kilmer, T.Kobilarcik,beams division, operators, MAB, SiDET …
July 21, 2000 G. Chiodini - Fermilab 14
VME COMPUTE
R
I/OBOARD STAR TFIB
ETHERNET
Setup: SSD telescope and DAQ
Magnet
DAQ
• I/O Board - Controls and Initializes 4 pixel readout chips - GPIB controller: HV, thresholds and pulser• Automatic pixel calibration
July 21, 2000 G. Chiodini - Fermilab 15
Setup: Pixel planes tested
FPIX0 64x12cells
8 bit external ADCFPIX1 160x18cells
2 bit internal FADC
•ST1-CiS p-stop•ST2-CiS p-sprayBonded active area 3.2x4.4mm2
•Two ST1-Seiko p-stop•ST2-Seiko p-sprayBonded active area 8x6.8mm2
Samples of FPIX0 and FPIX1 readout chips bump-bonded to ATLAS sensor prototypes
50x400 m2 cellsindium bump-bonding
n+np
July 21, 2000 G. Chiodini - Fermilab 16
Setup: Pixel planes tested
FPIX0Inner Board
Pixel detectorLight protected
Analog Buffer
8 bit ADC
FPIX1Inner Board
PC Board Interface
July 21, 2000 G. Chiodini - Fermilab 17
Setup: Pixel planes tested
Close-up middle station
Slots at different angles : 0, 5 10, 15, 20, 30 degrees
July 21, 2000 G. Chiodini - Fermilab 18
Results:pixel calibration - pulse generator
Qth=2500e-
Qnoise=106e-
Qnoise,ADC=400e-
Dynamic range MIP
FPIX0 bump-bonded to ST1 CiS p-stop
July 21, 2000 G. Chiodini - Fermilab 19
Results: Pixel calibration - X ray sources
Vth0 = Threshold knob
July 21, 2000 G. Chiodini - Fermilab 20
Results: Charge collection Single chip CiS p-spray
<Q>=21500e-
Qmp=18300e-
400 m
50 m
Charge losses are thought not to be intrinsic to the p-spray technology but a feature of this particular sensor design.
Charge losses
July 21, 2000 G. Chiodini - Fermilab 21
Results: Charge collection Single chip CiS p-stop
<Q>=30100e-
Qmp=24700e-
• The Landau distribution convoluted with a Gaussian function fit well the charge distribution.
• Only less than 0.7 % of the events have a signal less than 15000 e-.
)'
(
2')(
0
2
2
)'(2
2
MP
g
EE EEe
dENEfg
Saturation bump for CS=1
July 21, 2000 G. Chiodini - Fermilab 22
Results: Charge collection Single chip CiS p-stop
Landau distributionSingle pixel
Qmp
Qmp
July 21, 2000 G. Chiodini - Fermilab 23
Results: charge-sharing
Angle
[degs]
CS=1 CS=2 CS=3 CS=4 CS=5 CS6
0 .639 .328 .017 .009 .0034 .0025
5 .433 .527 .022 .010 .0041 .0028
10 .090 .846 .040 .015 .0055 .0029
15 - .635 .332 .022 .0080 .0034
20 - .209 .741 .031 .0124 .0060
30 - - .178 .769 .041 .0115
FPIX0 CiS p-stopQth=2500e- Vbias=-140V Vdep=-85V
Track inclinationDiffusion
Delta rays emission
c)
Relative fraction of cluster sizes (CS)
July 21, 2000 G. Chiodini - Fermilab 24
Results: charge-sharing Delta rays emission
July 21, 2000 G. Chiodini - Fermilab 25
Results: spatial resolutionPosition Finding Algorithm
Digital algorithm
x
Q Charge fluctuations
qLqR
Head-tail algorithm
Charge Sharing
2,1
,1 RL
Nj
Njj
digital
xx
j
x
x
)(_ fxx digitaltailhead LR
LR
where
Track position is correlated to the charge of the left and right hit in the cluster
LRtailhead q
qpitch
It reduces to charge-weighting for N=2 and f=pitch/2
22
RLLR qqq
qqq
nfluctuatio-qnoise
12
fraction Sharingpitchrmsdigital
July 21, 2000 G. Chiodini - Fermilab 26
Results: spatial resolutiondistribution and correlations
July 21, 2000 G. Chiodini - Fermilab 27
Results: spatial resolutionEta function f
July 21, 2000 G. Chiodini - Fermilab 28
Results: spatial resolutionGaussian fit residual distribution
ST1 CiS FPIX0 detector
• Xpred = projection of the kalman fit on the plane using all the planes, BUT the one under test pred = 2.1 m) .• Xmeas= coordinate measured by the plane under test using the head-tail analog interpolation.
CS=1,…,6
xy
400m
50m
Analog Pulse height used
July 21, 2000 G. Chiodini - Fermilab 29
Results: spatial resolutionSpatial resolution vs angle
ST1 CiS FPIX0 detector
Excellent spatial resolution at all angles using analog information
Qth= 2500e- Vbias= 140V Vdep= 85V
July 21, 2000 G. Chiodini - Fermilab 30
Results: spatial resolutionComparison between detectors
Qth=2500e-
Qth=2200e-
Qth=3780e-
• Most of the difference in spatial resolution between FPIX0 (nominal 8 bit) and FPIX1(2 bit) is due to the different readout threshold.• The charge losses in FPIX0 p-spray degrades the spatial resolution• BTeV requirement: better than 9 m
July 21, 2000 G. Chiodini - Fermilab 31
Results: spatial resolutionComparison with simulation
Comparison FPIX0 beam test data and simulation for binary and 8 bit analog readout
Comparison FPIX1 beam test data and simulation for 2 bit analog readout and 2 values of threshold
Good agreement between data and BTeV pixel detector simulation package with input parameters describing the detector
properties (such as Vbias, Vdep, …) corresponding to the sensors used in the test beam
July 21, 2000 G. Chiodini - Fermilab 32
Results: spatial resolutionComparison 8-bit ADC and 2-bit ADC
Comparison nominal 8-bit FPIX0 and “2-bit” FPIX0 degraded by software
FPIX0(8-bit) Qth=3720e-
FPIX1(2-bit) Qth=3780e-
Going from nominal 8-bit to 2-bit analog information the resolution degrade by less than 1m
FPIX0(8-bit) Qth=2500e-
July 21, 2000 G. Chiodini - Fermilab 33
Results: spatial resolutionBias voltage
ST1 CiS FPIX0 detector
For track angle > 5 degrees no degradation of the resolution when the detector is over depleted.
Nominal bias voltage
July 21, 2000 G. Chiodini - Fermilab 34
Results: spatial resolutionThreshold
ST1 CiS FPIX0 detector
As expected, larger readout threshold degrades the spatial resolution.
Nominal threshold
July 21, 2000 G. Chiodini - Fermilab 35
Results: spatial resolution2D spatial resolution
ST1 CiS FPIX0 detector
xy
Sigma=4.650.10 m
400+400 m
Good spatial resolution also when the charge isshared between the long pixel dimension.
July 21, 2000 G. Chiodini - Fermilab 36
Results: spatial resolutionNon-Gaussian resolution function
offcutpl
offcutoffcut
pl
pl
rxx
A
rxr
A
xF
||||
||||
)(
W
xt
W
bgpW
pWbg e
AdtxF
2
2)(
2)(
CS=1 and track angle < 10 degs:Square convoluted with a Gaussian
CS>1 track angle < 10 degsand all CS track angle > 10 degs:Gaussian + power law
Non-Gaussian part: •15% of events: half in the constant term and half in the tails• power law with an exponent 2
)()()( xFxFxF plGauss
July 21, 2000 G. Chiodini - Fermilab 37
Results: Magnetic field measurementFPIX0 p-stop in the fringe field
Fringe field up to 0.6T
Charge collection in magnetic field
Ratio double/singleVbias=-190V
geometry
Lorentz=,eB
July 21, 2000 G. Chiodini - Fermilab 38
Interaction vertex in the target
Interaction vertex in pixel plane
2.2mm thick
diamond target
Thousandsof triggeredmultipleinteractionsevents
Results: 4 plane pixel telescope
Excellent tracking capability even in high track enviroment
1mm
1mm
July 21, 2000 G. Chiodini - Fermilab 39
Further studies: Radiation Hard FE• Radiation dose in BTeV near the beam comparable to ATLAS first layer (1014cm-2y-1)• Pre-FPIX2 prototypes implemented in commercial 0.25m CMOS process from two vendors• Pre-FPIX2 irradiation tests with Co60 at Argonne confirm the RD49 Collaboration results at CERN
preFPIX2T irradiated at 33Mrad equivalent to 10 years running for BTeV at full luminosity of 21032cm-2s-1
ALICE/LHCb-RICH chip in DSM irradiated up to 30 Mrad (,p): no large Single Event Upset rate and no evidence of Gate Rupture Failure
July 21, 2000 G. Chiodini - Fermilab 40
Further studies: Module bench test
ATLAS 16 chips T1 p-stop
5 Fpix1 chips
HDI flex circuit
Layer Pair 1
Layer pair 2 Conductor
Dielectric Cu / Ni / Au
L AYER PAIR
Upilex-SGA
M1
M2
M3
M4
Multilayer Kapton High Density Interconnect cable. Very high density routing design:• Line center spacing = 40 m• Via center spacing = 208m (350m)
July 21, 2000 G. Chiodini - Fermilab 41
Shingled detectorNonporous carbon tubes, “flocking” carbon, and “fuzzy” carbon
Further studies: Mechanics and cooling
Heat exchanger test heated up by two aluminum plates
July 21, 2000 G. Chiodini - Fermilab 42
Conclusions• The FPIX-type FE performs well as
expected and needed• GREAT data sample to gain operational
experience with pixel silicon detectors (3M useful events)
• Primary features of the beam test results are clear:– Very good resolution at all angles– 3 bit ADC good choice for FPIX2– Little sensitivity to the bias voltage– Excellent tracking capability
• Good agreement between simulation and real data indicates a good understanding of the detector performance
• Rapid progress in global systems issue: bump-bonding, rad-hard sensor, rad hard front-end, modules, cooling, mechanical support, …