Operation of the CDF Silicon Detector
10th INTERNATIONAL CONFERENCE
ON INSTRUMENTATION FOR COLLIDING BEAM PHYSICS
Budker Institute of Nuclear Physics,
Siberian Branch of Russian Academy of Science,
Novosibirsk, Russia
February 28 - March 5, 2008
Silicon Detectors at CDF
 At the core of the CDF detector
 Largest operating silicon detector
 7-8 concentric layers of silicon
 7 m2 of silicon with 1.2 cm < r < 32 cm
 722,432 cha., 5644 chips, 704 sensors
 Designed only for Run IIa (~2/3fb-1)
 Upgrade for Run IIb was cancelled!
Silicon will have to survive through Run IIb (6/8 fb-1)
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
2
Silicon Sub-detectors
 Three Sub-detectors
 SVX II: 5 double sided layers
 Intermediate Silicon Layers (ISL): 3 double sided layers
 Layer 00 (L00): Single sided, LHC-style sensors
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
3
Sub-detector: SVX II
 SVX II: The core of the silicon systems
 Overall dimensions:
SVX II before installation
 1 meter along beam direction
 Radii from 2.5 to 10.6 cm
 Structure: three identical barrels
 2 bulkheads
 12 wedges
 5 concentric silicon layers
 Silicon layers
 Strip pitch: 60 to 140 mm
 Layers 0,1 and 3
(Axial and 90º strips)
 Layers 2 and 4
(axial and 1.2º strips)
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
SVX II barrel
4
Sub-detector: L00
 Layer 00: Right onto the beam pipe
 Overall dimensions:
 1 meter along beam direction
 Radii from 1.2 to 2.1 cm
 Structure: one single layer
 2 bulkheads
– Three consecutive sensors
 6 wedges
L00 during installation
 Silicon layers
 Strip pitch: 25 mm
 Axial strips
 Radiation tolerant (LHC style)
1.2 cm
2.1 cm
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
5
Sub-detector: ISL
 ISL: Intermediate Silicon Layers
 Overall dimensions:
 1.9 m along beam direction
 Radii from 20.5 to 29 cm
 Structure: three different barrels
 Central:
Single layer,14 wedges
 Forward:
Two layers,12 and 18 wedges
 Each barrel two bulkheads
ISL sensors
 Silicon layers
 Strip pitch: 55 mm
 Axial and 1.2º strips
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
6
Operational Issues
During commissioning:
 Blocked Cooling lines
 Blocked by glue, well inside the detector
 Solution: open them up with a powerful laser
 Resonances
 Wire bonds  to the magnetic field
 Synch. Readout  wire oscillate and break
 Solution: Stop high frequency synchronous
readouts.
 Beam Incidents
B
Jumper
groove left by beam
 High dose accidentally delivered to the detector
 Solution:
 Collimators in key parts of the Tevatron
 New Diamond based BLM system.
After commissioning: Infrastructure & Aging
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
7
Infrastructure & Aging: Power Supplies
 Common failure modes of CAEN SY527
 Communication loss
 Corrupted read back of voltages/currents
 Spontaneous switch off
 Failure mode of power supply modules:
 Voltages in Analog, Digital and Port-Card
supply start slowly dropping.
 Up to 47 Power supplies started to show this.
 Solution:
 Problem:
 aging of one type of capacitor
 36 capacitors per power supply
 Can result in bit errors
 Wait for the shutdown of September 2007 and …
 take all faulty power supplies out
 replace all 36 capacitors (on FNAL site)
 put them back in and test them on location.
 Time intensive effort, lasted about 2 months.
 Not enough time to change all
 Still expect to replace others as failure appears
All power supplies with this failure were replaced!
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
8
Infrastructure & Aging: Cooling Lines
 Cooling Lines
 Symptoms: electronic-valves start
failing.
 Problem: ISL cooling line (10% glycol
in water) became ACIDIC (ph=2)
during the 2006 shutdown
 Solution: coolant neutralized by
draining and larger use of de-ionizing
resin bed
 Ion chromatography
analysis showed
carboxylic acids, mostly
formic acid.
 Likely came from the
oxidation of glycol
 Welds of the aluminum rings that cool optical
transmitter had already been corroded
 One meter from the closest accessible point
 Why there ?
 Corrosion-resistance: is alloy-dependent
 Heat affected zone around junctions
manifold most sensitive (alloy: 6061-Al).
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
9
Infrastructure & Aging: Cooling Lines Repair
 Started shutdown of 2007:
 Keep the silicon cold and dry at all times
 A plastic tent was setup to work.
 A custom made air dryer changed the volume every 2 minutes.
 Dew Point was always kept below -10 Cº.
 Basic Idea:
 Cover holes with epoxy from
the inside of the pipe
 using borescopes and catheters.
Hole
 Repairs took a month
 4 shifts of people
 Current tests:
 Tight vacuum in the repaired lines
Repaired cooling system has been running stable for months !
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
10
Environmental Effects on Silicon Sensors
 Radiation effects:
 Modifies the crystal structure of the sensor bulk.
 Intrinsic parameters change with time.
 P-N junction evolves, and eventually disappears with time.
 Annealing effects:
 Due to temperature change of the sensors.
 Defects created by radiation are strongly affected.
 Performance of the sensor degrading with time: AGING
 Depletion voltage increases with time.
 Sensor has a maximum breakdown voltage.
 Can we fully deplete the sensor until the last day of operation ?
 Signal decreases with time and noise increases with time
 Can we keep good signal and noise levels until the last day of operation ?
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
11
Radiation Field
 Measured using more than 1000 thermo-luminescent dosimeters (TLDs)
 Two different data-taking periods allowed for distinction between fields:
due to beam losses
due to pp collisions
p
p
(See R. J. Tesarek et al., IEEE NSS 2003)
 Radiation field is collision-dominated and scales with
r  ( z ) , with 1.5   (z)  2.1
How this field affects the silicon sensors ?
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
12
Depletion Voltage: Signal Vs Bias
 Look at the charge collection distribution
 Reconstruct a track w/o using the studied sensor
 If track points to hit in sensor record its charge
 Charge collection distribution
 Follows a landau distribution
 Distribution is smeared by intrinsic noise
 Fit the curve to a Landau convoluted by Gaussian
(4 parameter fit)
 Depletion Voltage:
 Maximal for a fully depleted sensor
 Study charge collection as function of VBIAS
 Identify charge of Most Probably Value (MPV)
in each distribution
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
13
Depletion Voltage: Signal Vs. Bias
 Plot charge’s Most Probable Value for different bias voltages
 Fit to a sigmoid (parameters include the plateau of maximum charge)
 Define depletion voltage Vd
 Our criteria: voltage that collects 95% of the charge at the plateau
 Depletion Voltage as a function of luminosity
 3rd order polynomial fit around the inversion point
 Linear fit to extrapolate to the future
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
14
Depletion Voltage: Noise Vs Bias
 Take advantage of double sided sensors, that have strips on the back side
 Depletion zone grows from the p+ side
 Noise on the other side’s strips (n+) reduce when the depletion zone reaches them.

Need a criteria for defining Vd

No beam required

Does not work after the sensor underwent inversion.
 We use 95% reduction in noise between the two plateaus
 no interference with data-taking
 Depletion zone generated differently
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
15
Depletion Voltage: Results
 Prediction for L00
 Prediction for SVX-L0
 Depends on type of sensor
 Oxygenated ladders invert much later
We should be able to deplete sensors until the end of Run II
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
16
Signal to Noise Ratio
 The figure of merit of the performance is the Signal to Noise Ratio (S/N)
 Signal: charge collected when a charged particle crossed the sensor
 Noise: intrinsic noise of the detector
 Mean Strip Noise
 Signal
 Use J/y  m+m- tracks
 Get total charge of cluster
 Decrease linearly with Lum.
Feb 29th, 2008
 Average over strips in charge cluster
 Obtained from calibrations taken
every two week.
 Square root increase with Lum.
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
17
Signal to Noise Ratio
 Fit of S/N
 Limit I: S/N=8 (SVT eff.)
 S/N = 6, ~5% loss in SVT eff.
 Limit II: S/N=3 (B tag eff.)
 Sensor-type behavior
 Layers 2,4 (Micron)
 Layers 0,1,3 (Hamamatsu)
 First layer
 careful monitoring to see if it is
going to be useful at 4/5 fb-1.
Most of the silicon layers will be fully operational until the end of Run II
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
18
Conclusions
 Operational Issues:
 Advice : expect the unexpected.
 We have recovered cooling to the full detector subsystems.
 Power supply modules with unstable voltages fixed at FNAL
 Sensor Aging
 Data indicates that we will be able to fully deplete the sensors
 The innermost layers of the detector have passed type inversion
 Studies indicate we will remain with a very effective S/N ratio.
The CDF Run II Silicon Detector will continue successful
operation for the rest of Run II !!
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
19
Backup Slides
Radiation Effects on Silicon Sensor
 Bias current increases with integrated radiation
 Theory:
fluence
Ileak  α Φ V
sensor’s volume
damage factor
 Can be used to obtain flux
 Comparison with TLD’s prediction
Flux 
Φ
 dL

I leak
αV  dL
(See R. J. Tesarek et al., IEEE NSS 2003)
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
21
Fermilab and the Tevatron
 Fermilab (1967)
Fermi National Accelerator Laboratory – Aerial View
 Large number of H.E.P. projects
Tevatron
 Tevatron Run II (2001–2009)
2 km
 Proton-antiproton collider
 Two multi-purpose detectors
 CDF & DØ
[Fermilab Visual Media Service]
 Proton-antiproton collider
 √s = 1.96 TeV, 36×36 bunches
 Record instant. peak luminosity
292 µb–1 s–1 (1 µb–1 s–1  10–30 cm–2 s–1)
 Expect 6–8 fb–1 by end of Run II
Feb 29th, 2008
Ricardo Eusebi - INSTR 08, Novosibirsk, Russia
22