Thoughts on Disk and Barrel Silicon Tracker Geometry

1
Thoughts on Disk and Barrel Silicon Tracker
Geometry

• (A Working Proposal)
• W. E. Cooper, M. Demarteau, M. Hrycyk
• Fermilab

2
General Comments

• Of the two designs previously suggested by Marty
  Breidenbach, we have begun to investigate the
  design with concentric barrels whose length
  increases with radius. In terms of mechanical
  fabrication and support, we think such a design
  may lead to less material and more favorably
  oriented material than one with all, or multiple,
  barrels of the same length.
• We note that ladders of long barrels are awkward
  to build and that disks which join barrels can be
  used to improve the barrels out-of-round
  stiffness.
• For tracking in a solenoidal field, barrel and
  disk geometries exhibit obvious and significant
  differences.
• A barrel gives a hit for which R is automatically
  known with reasonable precision. To reconstruct
  trajectories in the X-Y plane (beam parallel to
  the Z-axis), only Phi needs to be measured.
  Knowing the sensor gives R and a range of Z.
• A disk gives a hit for which Z is automatically
  known with reasonable precision (maybe
  unnecessarily well known). To reconstruct
  trajectories in the X-Y plane, both R and Phi
  need to be measured.
• If occupancies are low enough that additional
  stereo planes are unnecessary, then, with
  single-sided sensors, disks represent nearly
  twice the material thickness as do barrels.
• We have assumed that power dissipation is low
  enough to allow cooling by forced air convection.

3
Overall Geometry

• Barrels
• Five barrels
• Measure Phi only
• Eighty-fold phi segmentation
• Barrel lengths increase with radius
• Maximum active radius 1.240 m
• Minimum active radius 0.218 m
• Maximum active length 3.307 m

• Disks
• Five double-disks per end
• Measure R and Phi
• Disk radii increase with Z
• Maximum active radius 1.262 m
• Minimum active radius 0.041 m
• Maximum Z (active) 1.687 m
• Minimum Z (active) 0.282 m

4
Barrel Features

• 100 overlap of sensor active regions is provided
  in both Z and phi.
• Adjoining sensors in Z are offset radially by
  1.9 mm.
• Adjoining sensors in phi are offset radially by
  5 mm.
• Double-walled, CF-based support cylinders are
  shown.
• Separation of inner and outer walls is set by the
  required out-of-round stiffness and provided via
  spacers (Rohacell?).
• As a practical matter, the outer cylinder would
  probably be an 80-sided polygon with spacers
  (Rohacell?) under the B-layer sensors.
• The inner wall could be circular and could be cut
  shorter than the outer to allow space for disks
  and mechanical connections.

5
Barrel Sensors

• Five types of sensors with a single type per
  layer
• All sensors are assumed to be single-sided and to
  have traces parallel to the beam line.
• Could add stereo layers by widening and rotating
  sensors
• Sensor design is straight-forward and
  conventional.
• Pitches of 50 µm and 51 µm
• Could be varied without complications
• Sensors could have intermediate strips.
• Sensor cut lengths
• 103.6, 110.2, 112.4, 113.5, and 141.7 mm
• Sensor cut widths
• 21.2, 40.4, 59.6, 80.4, and 100.0 mm
• Barrel 1 through 3 sensors fit within 6 wafers.
• Barrel 4 and 5 sensors require 8 wafers.

6
Sample Barrel Layout (1)

• Tentative layouts have been made for all 5
  barrels.
• Overlap between sensor active regions 1.0 mm
  (0.5 mm per sensor) in both Z and phi.
• Guard ring boundary width 1.02 mm
• Consistent with recent sensors from HPK

7
Sample Barrel Layout (2)
8
Barrel Readout

• While not incorporated in the barrel layout
  shown, D0 Run 2b sensor / hybrid geometry may be
  applicable.
• Each double-ended hybrid serves two sensors and
  joins them into a sensor / hybrid module.
• One chip channel per sensor channel
• All sensor / hybrid modules of a barrel layer can
  be identical.
• But changes would be needed to provide full
  Z-overlap.
• Modules can be joined into staves, but
  temporary, external support is likely to be
  needed to handle staves of any significant length
  (gt 0.6 m).
• We assumed a 128-channel readout chip.
• Depending on barrel layer, 3, 6, 9, 12, or 15
  chips would be needed per sensor.
• That would be consistent with smaller numbers of
  wider chips, each with 384 channels.
• Optical signal connections from near the ends of
  barrels may be convenient, if devices with
  sufficiently low power dissipation become
  available.
• Grounding between sensors, hybrids, and any
  conductive support structure (such as one based
  upon carbon fiber) would be critical.

D0 Run 2b Axial Prototype
D0 Run 2b Stereo Prototype
9
Grounding

• D0 has obtained excellent results with copper
  mesh on kapton circuits co-laminated to carbon
  fiber structures.
• Extensive readout testing at Fermilab. CF
  structures provided by UW.
• Separate mesh circuits are laminated to the back
  surfaces of sensors.
• Circuits wrap around a sensor end (or side) to
  allow connections to be made on the top surface.
• Connections between the two sets of circuits are
  made when sensors are installed.
• The techniques are applicable to both barrels and
  disks.

Prototype CF cylinder from the University of
Washington with laminated copper mesh showing
contact regions for connections.
Test connection from CF to dummy sensor. R 0.2
O.
10
Noise Estimate

• Consider four dominant contributions to noise
• shot noise from detector current
• parallel noise from Al on strips
• thermal noise from series bias resistors
• Front-end noise
  Q ( 300 41 CL(pF) )

• Consider four dominant contributions to noise
• shot noise from detector current
• parallel noise from Al on strips
• thermal noise from series bias resistors
• Front-end noise

ENC for innermost layer, smallest contribution
from Ileak
11
Signal to Noise Ratio

• Compare S/N ratio for
• innermost and outermost layer
• Two temperatures T20o C and T -10o C
• Note based on ENC for simple CR-RC circuit
  scaling of FE noise over large range of shaping
  times may be oversimplified

12
Disk Layout

• The layout is more difficult and less completely
  developed than that of barrels.
• There are still some difficulties with the
  layouts of disks 1A and 1B.
• We followed the same general assumptions as were
  applied to the barrels
• 1.0 mm overlap between sensor active regions
• 1.02 mm wide border for guard rings.
• Sensors are on the outward-facing surfaces of
  disks.
• Adjacent sensors are positioned at slightly
  different Zs to provide R and Phi overlap.
• A-disks fit a short distance within barrels and
  are supported from B-disks.
• The populated surface of a B-disk covers the end
  of a barrel. A mechanical connection is made
  from the unpopulated surface of the disk to the
  end of the barrel.
• The unpopulated portion of the disk extends to
  the inner radius of the next larger barrel, where
  mechanical connections are made.

13
Disk to Barrel Connections

• Disk to barrel connections are critical in this
  design. One end is shown.
• At the other end, the stud is attached to the
  disk in order to allow the connection to be
  completed. Barrels are clocked during
  installation to allow teeth to pass one another.

14
Sample Disk Layout (1)

• Preliminary layouts have been developed for all
  five disks. Wafers of 6 and 8 technologies
  have been assumed. Disks 4 and 5 are shown in
  the table.

15
Sample Disk Layout (2)

• Wedge cut, rather than active, edges were aligned
  at a constant phi.
• That allows alignment of wedges using cut edges,
  a significant simplification during assembly.
• For a constant overlap distance, traces do not
  quite point towards the beam line.
• The eighty-fold symmetry at the outer disk radius
  (disk 5) matches that of barrels.
• To maintain sensible trace lengths, the phi
  multiplicity with which sensors are arrayed has
  been decreased as one moves to smaller radius
  80, 40, 20-fold in disk 5 and as low as 10,
  5-fold in disks closer to Z 0.

Disk 5B, Y sensors
Disk 5A, Phi sensors
16
Wedge Sensors and Readout

• In the Y sensors, traces run parallel to the
  inner wedge edge and have constant pitch.
• Trace lengths vary.
• Hybrids have readout chips arrayed parallel to Y.
• Transverse position of the hybrid is arbitrary
  within reasonable limits.

• In the Phi sensors, trace ends are evenly spaced
  laterally at inner and outer edges. The pitch
  increases linearly with Y.
• Variations in trace length are modest.
• Hybrids have readout chips arrayed parallel to
  the inner wedge edge.
• Y position of the hybrid is adjusted to match
  chip pitch to sensor pitch.

Disk 5B, Y sensor 4
Disk 5A, Phi sensor 4
17
Pedestal Variation Across Wedge

• Pedestals will vary across wedge and with ring
• Trace length varies
• Sensor area increases
• Estimate variation for phi- and r-wedges for each
  disk separately

18
Sensor and Chip Counts

• Chip counts are high, but that is consistent with
  a pitches of 50 - 60 µm, trace lengths of 100
  140 mm, and no ganging of sensors.
• We should examine what is driving sensor pitch
  and how pitch should vary with R and Z.
• Momentum resolution? Occupancy? (Should not be
  vertex resolution).
• Associated heat load is a concern for air
  cooling.
• We think 0.05 W/chip is a reasonable estimate,
  but that leads to an overall heat load of 6373
  Watts.

19
Servicing Trackers

• How many elements need to be removed to service a
  particular device (barrel, disk, VTX)?
• With barrel-to-barrel connections an integral
  part of disks, all disk/barrel sets outside a
  particular barrel (or disk) would need to be
  removed to reach that barrel (or disk).
• Modifying connections between barrel layers so
  they are separate rings would allow disks to be
  removed while barrels remain.
• Readout choices (cables, optical fibers,
  wireless) have a major impact on the space needed
  to provide services and impact the frequency with
  which service may be needed.

20
Other Mechanical Design Issues

• Clear volume to be preserved for VTX
• Paths for services to and from VTX
• Support of VTX from SiD
• External support of SiD
• How sure are we that no additional stereo layers
  are necessary?
• Our naïve guess would have been that stereo would
  be needed for the two or three barrel and disk
  layers closest to Z 0.
• Is the angle for the transition from barrel to
  disk geometry optimum?
• Should the transition angle be smaller, say 30o?
• What is an appropriate trade-off between 100
  coverage with an individual barrel or disk and
  the complications from overlapping sensors?
• Should disk phi traces should be truly radial?
• Should disk Y traces instead be R traces?
• A suggestion Establish and maintain a web page
  with current design parameters (unless there
  already is one).

21
In Conclusion

• We have begun looking at one design based upon
  sensors of dimensions which should be obtainable.
• Barrel and disk layouts are reasonably matched.
• Initial sensor counts and dimensions have been
  given.
• Barrel hybrids are consistent with 384 channel
  readout chips.
• Disk hybrids are consistent with 512 channel
  readout chips.
• Chip counts are high.
• Power dissipation is a concern.
• Concepts for support structures, based upon known
  technologies, have been suggested.
• We would imagine updating with new parameters as
  studies are done and also adding details of
  hybrid structures and services.
• At that time, sufficient information should be
  available for a realistic estimate of the tracker
  thickness in radiation lengths.
• We would be glad to continue with this design and
  to begin to examine other designs and design
  options.

22
Back-up Slides D0 CFT

• Central Fiber Tracker barrels are based upon
  scintillating fiber ribbons supported upon
  CF-Rohacell-CF double-walled cylinders.
• CF rings join the barrels.
• The six outer barrels of the tracker are 2.54 m
  long. The inner two barrels are 1.68 m long.
  Radii at which fibers are placed range from 0.20
  to 0.52 m.
• The silicon tracker is supported from the
  innermost CFT barrel.

23
Back-up Slides D0 Run 2b Central Tracker

• Silicon within CFT