1 The BaBar Silicon Vertex Tracker (SVT) Claudio Campagnari University of California Santa Barbara
Jan 05, 2016
1
The BaBar Silicon Vertex Tracker (SVT)
Claudio CampagnariUniversity of California
Santa Barbara
2
Outline
• Requirements
• Detector Description
• Performance
• Radiation
3
SVT Design Requirements (TDR)Performance Requirements• z resolution < 130 m
• Single vertex resolution < 80 m.
• Stand-alone tracking for PT < 100
MeV/c.
Performance Requirements• z resolution < 130 m
• Single vertex resolution < 80 m.
• Stand-alone tracking for PT < 100
MeV/c.
PEP-II Constraints• Permanent dipole (B1) magnets at +/- 20 cm from IP.
• Polar angle restriction: 17.20 < < 1500.• Must be clam-shelled into place after installation of B1 magnets
• Bunch crossing period: 4.2 ns (nearly continuous interactions).
• Radiation exposure at innermost layer (nominal background level):• Average: 33 kRad/year.• In beam plane: 240 kRad/year.
• SVT is designed to function in up to 10 X nominal background.
PEP-II Constraints• Permanent dipole (B1) magnets at +/- 20 cm from IP.
• Polar angle restriction: 17.20 < < 1500.• Must be clam-shelled into place after installation of B1 magnets
• Bunch crossing period: 4.2 ns (nearly continuous interactions).
• Radiation exposure at innermost layer (nominal background level):• Average: 33 kRad/year.• In beam plane: 240 kRad/year.
• SVT is designed to function in up to 10 X nominal background.
4
SVT characteristics• Five layers, double sided (R and z)
– Barrel design, L4 and 5 not cylindrical– 340 wafers, 6 different types– Low mass Kevlar-Carbon Fiber support ribs
• Upilex fanouts to route signal to ends• Double-sided AlN HDI (104 of these)
– Outside tracking volume– Mounted on Carbon Fiber cones (on B1
magnets)• Atom chips
– 1156 chips, 140K channels
5
Space Frame and Support Cones…mounted on B1
magnets
6Layer 1,2,(3): vertexingLayer (3),4,5: tracking
7
SVT Modules
Z-SideZ-Side
Phi-SidePhi-Side
Si Wafers Carbon/Kevlar fiberSupport ribs
High DensityInterconnect
(mechanical model) Micro-bondsMicro-bondsFlexible Upilex Fanout
Fanout Properties:• < 0.03 % X0
• 0.52 pF/cm
Fanout Properties:• < 0.03 % X0
• 0.52 pF/cm
8
SVT High Density Interconnect
AToMChips
UpilexFanout
MountingButtons
BergConnector
Flexible Tail (testing version)Functions:
• Mounting and cooling for readout ICs.
• Mechanical mounting point for module.
Functions:
• Mounting and cooling for readout ICs.
• Mechanical mounting point for module.
Features:• AlN substrate.• Double sided.• Thermistor for temp. monitor.• 3 different models.
Features:• AlN substrate.• Double sided.• Thermistor for temp. monitor.• 3 different models.
9
Silicon Wafers
Features•Manufactured at Micron.•300 m thick.•6 different wafer designs.•n- bulk, 4-8 kcm.•AC coupling to strip implants.•Polysilicon Bias resistors on wafer, 5 M
Features•Manufactured at Micron.•300 m thick.•6 different wafer designs.•n- bulk, 4-8 kcm.•AC coupling to strip implants.•Polysilicon Bias resistors on wafer, 5 M
Bulk Properties•Bias current: 0.1 to 2.0 A
•Bulk current: 0.1 to 2.0 A
•Depletion voltage: 10 to 45 V
Bulk Properties•Bias current: 0.1 to 2.0 A
•Bulk current: 0.1 to 2.0 A
•Depletion voltage: 10 to 45 V
Strip Propertiesn-side n-side n-side p-side
•Strip Pitch: 50 m55 m 105 m 50 m
•Inter-strip C: 1.1 pF/cm 1.0 pF/cm 1.0 pF/cm 1.1 pF/cm
•AC decoupling C: 20 pF/cm 22 pF/cm 34 pF/cm 43 pF/cm
•Implant-to-back C: 0.19 pF/cm 0.36 pF/cm 0.17 pF/cm
•Bias R: 4 to 8 M 4 to 8 M 4 to 8 M 4 to 8 M
Strip Propertiesn-side n-side n-side p-side
•Strip Pitch: 50 m55 m 105 m 50 m
•Inter-strip C: 1.1 pF/cm 1.0 pF/cm 1.0 pF/cm 1.1 pF/cm
•AC decoupling C: 20 pF/cm 22 pF/cm 34 pF/cm 43 pF/cm
•Implant-to-back C: 0.19 pF/cm 0.36 pF/cm 0.17 pF/cm
•Bias R: 4 to 8 M 4 to 8 M 4 to 8 M 4 to 8 M
10
Silicon Wafers
Bias ring p+ Implant
Al
p+ strip side
P-stop n+ Implant
Polysiliconbias resistor
Polysiliconbias resistor
Edge guard ring
Edge guard ring
n+ strip side
50 m
55 m
11
The AToM Chip
Features:•128 Channels per chip•Rad-Hard CMOS process (Honeywell)•Simultaneous
– Acquisition– Digitization– Readout
•Sparsified readout•Time Over Threshold (TOT) readout•Internal charge injection
Features:•128 Channels per chip•Rad-Hard CMOS process (Honeywell)•Simultaneous
– Acquisition– Digitization– Readout
•Sparsified readout•Time Over Threshold (TOT) readout•Internal charge injection
Custom Si readout ICAToM = A Time Over threshold Machine
Custom Si readout ICAToM = A Time Over threshold Machine 5.7 mm
8.3 m
m
12
The AToM Chip
CAL DAC
Shaper
ThreshDAC
CompPREAMP
TOT CounterTime Stamp
Event TimeEvent Number
Revolving Buffer
193 Bins
Si
Buffer
Buffer
Chan # Sp
ars
ifica
tion
Re
ad
ou
t Bu
ffer
CINJ
CAC
SerialData Out
Amp, Shape, Discr, Calib•5-bit CAL DAC (0.5 fC/count)•5-bit Thr DAC (0.05 fC/count)•Shaping time 100 - 400 ns•Typical threshold 0.6-0.9 fC
Amp, Shape, Discr, Calib•5-bit CAL DAC (0.5 fC/count)•5-bit Thr DAC (0.05 fC/count)•Shaping time 100 - 400 ns•Typical threshold 0.6-0.9 fC
Trigger Latency Buffer•15 MHz Sample rate•Total storage = 12.7 us
Trigger Latency Buffer•15 MHz Sample rate•Total storage = 12.7 us
TOT, Tstamp, Buffering•4 bits TOT (logarithmic)•5 bits Hit Tstamp
(67 ns/count)•4 buffers / channel
TOT, Tstamp, Buffering•4 bits TOT (logarithmic)•5 bits Hit Tstamp
(67 ns/count)•4 buffers / channel
15 MHz
13
Performance
• Calibration, Noise• Occupancy• Efficiency• Intrinsic Resolution
14
Calibration• Noise, gain, pedestals, bad
channels obtained from scanning threshold with and without charge injection and counting hits– 600K errfun fits, 150K linear fits– once a day; takes ~ 2 minutes
• Very stable• Downloadable chip parameters
have not changed between Oct 1999 and ~ 2004 – needed to change because of rad
damage
Threshold
Hits
Offset
Noise
Qinj CountsO
ffset
Cou
nts
Threshold DACOffset
15
Alignment: a curiosity• SVT tied to machine elements, not to DCH• SVT is always moving w.r.t. BaBar due to e.g.
thermal excursions.• Position of SVT as rigid body is monitored
and fed back to reconstruction ~ every hour
16
NoiseLayer ENC Layer ENC
1 1200 1z 880
2 1240 2z 970
3 1440 3z 1180
4 1350 4z 1210
5 1600 5z 1200
Layer ENC Layer ENC
1 1200 1z 880
2 1240 2z 970
3 1440 3z 1180
4 1350 4z 1210
5 1600 5z 1200
1 MIP at normal incidence, about 23,000 electrons
17
Cluster efficiency
(SW + HW)
Excluding malfunctioning readout sections
18
Resolution
Blue: dataRed: MC
19
Standalone reconstruction of low PT tracks
Reconstruction of s fromD* D s
is (mostly) with SVT alone
20
Most of the detector is working• Redundancy built in, e.g., 2 data and control paths• Chips are only active electronics that is not accessible • Layer 1 perfect• 4/208 sections not working• Problems are from
• shorts on hybrids• elec. probems on wafers
short
Both noisy
2 chips masked short
21
Radiation
• Monitored by 12 diodes at ~ radius of layer 1• Diodes can abort beam• Operation tricky due to heavy radiation damage• Now also use diamonds
SVTRAD System
22
Measured absorbed Dose
midplane
non-midplane
Few MRad in the horizontal planeMostly from showers from off-momentum beam particlesNot from Physics
23
Consequences of high doses
• Bulk damage to Si– increase ILEAK increase in noise
– type inversion
• Damage to chips– originally tested to 5 MRad with Co
source
24
Bulk Damage
from Moll
NIEL scaling: high energy electron causesignificant damage (~1/10 of hadrons)
Not appreciated by us originally
~ D
am
age e
ffect
iven
ess
25
• C-2 vs V curve show inversion• Results in ~ agreement with NIEL scaling hypothesis• Charge-collection-efficiency after local type inversion measured: OK
• C-2 vs V curve show inversion• Results in ~ agreement with NIEL scaling hypothesis• Charge-collection-efficiency after local type inversion measured: OK
Tests at Elettra (Trieste)
26
Expected Atom Chip Damage
Radiation damage on AToM chip• studied using Co60 sources at LBL and SLAC, up to 5 MRad.• No digital failures (if chip power on)• Gain loss 4.2% / MRad• Noise increase: 19%/MRad
27
HDI Card in horizontal plane
Th
resh
old
off
set
(cou
nts
)
Channel
Unexpected Phenomenon: Pedestal Shift
• Pedestal (Threshold offset) started to increase• Behavior associated with AToM chip location, not with
strip location• Remember: we have 1 pedestal value per chip!!!
Chip 4
20 threshold DACs = 1fC
Ped
est
al
28
Pedestal Shift (cont.)• Sets in at an integrated radiation dose of 1 Mrad• But then it recovers. • Effect reproduced at Elettra• Ride out the storm by adjusting thresholds as
well as we can
Effect reproduced @ Elettra
Channel
Del
ta T
hre
shol
d (
cou
nts
)
narrow e- beam AToM Chip
Pedestal recovers
Th
resh
old
off
set
(cou
nts
)
Integrated Radiation
1 Mrad
2 Mrad
Groups of 8 channels
29
Unexpected Phenomenon:Leakage Current Increase
• Since May 2004 an anomalous increase in the bias current for some modules has been observed
• Only Layer-4 modules: not a simple radiation damage effect
• No geometrical correlation • Consequences:
increasing occupancies• Coincides with beginning of
"trickle injection" operation– beam always on!!!!!
300uA
10uA
Apr May Jun
I Lea
k (
A)
Days in 2004
30
I Lea
k (A
)
Time (hrs)
Leakage Current Increase
-20V
-20V
+20V
+20V
E
Nside
Pside
Nside
Pside
VL5-L4
=+40V
Hypothesis:
Accumulation of static charge on the silicon surface. The charge is beam-induced drifts because of the field between the facing sides of different layers.Causes increase in electric field at junctionedge, inducing a soft junction breakdown.
By varying the potential drop across the air between the layers we can control the effect
1800 0000 0600 1200Solution: change relative voltages, L4 vs L5
Also, increase humidity of air
31
Conclusions
• BaBar SVT has been working well for about 7 years now– Installed and cabled in April 99– Taking physics quality data since June 99
• There have been a few surprises along the way, but we have managed to survive