Vertex Detector: Engineering Issues

Craig ButtarUniversity of Glasgow

Cambridge GLDC meetingSept 07

Design Features

• Outer radius ~ 6 cm• Barrel length ~ 14 cm• Ladder widths 1-2 cm• Disks to cover

forward region

(GLD)

(LDC)

(SID)

A bit larger than this

Optimizing Vertex Performance

• Close to IP – Reduce extrapolation error– Inner radius ~1.5cm

• Position resolution (<5 microns)– Impact parameter resolution

≤ 5µm 10µm/(p sin3/2 )• Minimise multiple scattering

– Material ~ 0.1X0/layer

• 5 m resolution or better is possible with current sensor technology– Need good alignment to exploit this

• Minimal mass is crucial– Constraints on mechanics– Constraints on power

• Cooling• Power delivery

– Alignment

Parametric simulation assuming:• 0.1% RL per layer• 5 micron resolution• 1.4 cm inner radiusVarying each parameter

IP Resolution, 1 GeV tracks

0

5

10

15

20

25

0 5 10 15 20 25

Hit resolution (micron) or RL x 10^-3 or IR (mm)

Mic

ron

s

varying resolution

varying radiation length

Varying inner radius

ILC target

Material budget

• Service handling at ends of barrel are the problem

• The boring stuff is important!

• Breakdown for pixels

cos=0.95 cos=0.95ATLAS Tracker

Sensors (300m) 1.1%

Ro+bump bonds 1.4%

Hybrid 1.0%

Local support+cooling 5.4%

Cables 0.3%

Global support 1.5%

Total for 3 layers 10.7%

Beam pipeCF supportepoxySilicon

Mechanical Support• 0.1% X0/layer 100m of Si

• Need to start with thin Si, typically 20m• Thin supports

– Carbon fiber-based supports, similar to D0 layer 0/CDF Layer00

– Foam-based (SiC, RVC) supports (LCFI)

– Silicon picture frame (MPI)

• System Issues– Planarity of the sensors

– Bonding to thin silicon

– Thermal bowing

– Connection to external cables

MPI Design

(University of Washington)

(LCFI)

(SID inside support cylinder)

SiC Foam Ladder

• 20 um thick silicon• 1.5 mm thick SiC foam

– 8% relative density• Silicone adhesive pads

– 1mm diameter 200 microns high on ~5mm pitch

• ~0.14% X0SiC ladder

ladder block

glue

annulus block

mm

um

mm

um

LCFI

RVC Foam/Silicon Sandwich Ladder

• 20 micron thick silicon• 1.5 mm thick RVC foam

– 3% relative density• Silicone adhesive pads

– on ~5mm pitch• Tension ~1.5 N• ~0.08% X0

RVC sandwiched ladder

silicon spacer

ladder block

glue

annulus block

Tension

umum

mm

mm

LCFI

Air Cooling

• Air cooling is crucial to keep mass to a minimum– Require laminar flow through available apertures– This sets total mass flow – other quantities follow– Implies a limit on power dissipation

• For SiD design – Use the outer support CF cylinder as manifold (15mm r)– Maintain laminar flow (Remax = 1800). – Total disk (30W) + barrel (20W) power = 50W average

• For SiD ~ 131 µW/mm2.• Max T ~ 8 deg

(Cooper, SID)

Cooling StudiesTest model of 1/4 Barrel• Cold nitrogen cooling• Heaters at ladder ends• Parallel CFD simulations

• Flow 5-20 SLM

– 0.52 g/s whole detector

– Laminar flow

Power Extracted (W)

Temperature Difference (K)

LCFI

Alignment is critical

• ILC physics programme depends on identification of secondary vertices

• Ability to do this depends on tracking resolution

• Tracking resolution dependent on alignment precision

• Individual hit resolution may be O(5) m– Alignment must be better, so that contribution in quadrature

does not degrade hit resolution

Alignment – LHCb VELO

Rigiditylow CTEoverlaps10m alignment

Hardware Design SoftwareMetrology

Measurement machineIndividual modules during assemblyComplete system10m alignment

Alignment at few m levelIterative / non -iterative methods

BEFORE / AFTER

For ILC vertex detectorPosition of detectors on ladders to ~10mThin detectors Warping (SLD)Thin ladders not rigidLow mass beam pipe Vertex detector will move wrt experiment

Design• Design into system features for alignment

– Rigidity, thermal and humidity expansion• This is difficult at low mass

– Overlaps – not just for coverage, e.g. – VELO left, right half overlap– SLD CCDs

Metrology - importance• Starting point for alignment parameters• Constrains degrees of freedom not accessible from

alignment system• e.g. large systematic on particle lifetimes is radius of barrel e.g. +/- 40

um on 4cm = 1%• e.g. aspect ratio of vertex detector gives systematic – important for FB

asymmetries

• Define/understand elements:

– Ladders• Ideally rigid, 6 dof/ladder (372 for LCFI barrel)• Ladders are not a rigid object eg detector bow, CTE

– Develop models? Difficult to measure during construction need to understand effect of thermal changes eg CTE, tension due to mechanics and services? (CTE studies by LCFI)

– Greater no. of degrees of freedom than ladders x 6 (ATLAS has 34,992 dof)

– Requires good initial survey and understanding of changes» Difficult to do under in-situ conditions

Power delivery

• High currents to drive CCD clock pulses• Minimise voltage drop on power cables

– Low resistance more conductor mass (Cu)– 0.5V drop at 6cm ~ 0.5%X0

• Use serial powering– Power at higher voltage, locally regulate at detector– Reduces conductor mass– 0.5V drop at 6cm ~ 0.04Xo

• Issues– Failure in string– Coherent noise– Increase complexity of interconnects

• UK-ATLAS activity for sLHC upgrade

UK Experience

• ATLAS barrel and endcap silicon tracker, LHCb VELO– Sensors (strips)– Readout electronics– Module construction– Engineering– Cooling – liquid based– Alignment

• LCFI• SLD CCD based vertex detector• ALEPH, DELPHI, OPAL strip-based vertex detectors• CDF Layer-00 strip-based vertex detector

Summary/conclusions• Low mass critical to achieve required IP

– Challenging eg ATLAS is ~ 10.7%X0 for 3 pixel layers– Dominated by support and cooling

• Target layer thickness 0.1%X0 (100m Si)– Thin sensors– New support materials– Air cooling limits power to ~O(10W)– Also implications for services serial powering

• Need to consider alignment in hardware– Design: overlaps in system (increase material)– Metrology during assembly– Warping of thin detectors and ladders– Report of LHC alignment workshop: CERN yellow report 2007-004

• Thanks to: Mark Thomson, Tim Greenshaw, Joel Goldstein, Chris Parkes, Val O’Shea, Richard Bates

Barrel Layout

Beam pipe Ladder (detector element)

Foam cryostat

Beryllium support shell

Spring

Ladder block

Annulus block

Fixed end Sliding end

Readout and drive chips

Substrate

Silicon sensor

Beryllium support shell

Annulus and ladder blocks

Barrel Layout

Layer no

No of Ladders

Radius(mm)

Active length(mm)

Active width(mm)

Tilt angle

Overlap(mm)

1 8 15(19) 100 13 0 0

2 8 26(28.5) 250 22 0 0.42

3 12 37 250 22 15 1.3

4 15 48 250 22 15 0.86

5 19 60 250 22 15 1.2

• Looking at:– the radius of the layers– width of elements– tilt angle

Metrology - Equipment• Smartscope

• Small scale items – not full system– High precision O(2) m XY O(10) m Z– Optical head– Automatic pattern recognition– Excellent for measuring sensor curvature– Individual sensors not double sided modules – no

alignment to reverse side

Residuals are function of the detector resolution and the misalignments

From this…

The geometry we are looking for is the one which minimizes the tracks residuals

… to that

Alignment principle :

Software Alignment

Each individual ‘unit’ has six degrees of freedomNeed to apply global transformation constraints

•All software alignment procedures follow one of these two forms:

conclusion : both methods can be made to work well.

Misaligned detector

Geometry not corrected

Plot and fit the residuals

distributions

Best mean and values ?

Detector aligned

YES

NO

Fit the tracks

Modify the

geometry

ITERATIVE

Misaligned detector

Geometry not corrected

Detector aligned

YES

NOFit the tracks & the residuals

Outliers rejected ? Non-linearities

corrected ?

NON-ITERATIVE

Iterative / Not Iterative

Iterative: fit biased tracks then fit alignment constants, iterate to reduce biasNon-Iterative: fit tracks and alignment constants simultaneously

① Establish linear expression of residuals as a function of mis-alignments.

Fit the tracks simultaneously with the alignment constants

Get all track parameters and all misalignment constants simultaneously

1 single system to solve. But this system is huge ! (Ntracks∙Nlocal+Nglobal

equations)

BUT…

xclus = xtrack + x

Parameters i of the tracks(different for each track)

xclus = ∑i∙i + x

LOCAL PART

xclus = ∑i∙i + ∑aj∙j

Residuals expressed as function of misalignments

i

GLOBAL PART

Global Alignment Method – H1, LHCb, ATLAS

rclus = (xclus - x)∂ 2

∂ ∆i

∂ 2

∂ i

= = 0

Alignment minimise 2res = ∑ ∑wclus∙r2

clus

The matrix to invert has a very special structure:

Inversion in section (implemented in the code MILLEPEDE V.Blobel - NIM. A 566), The problem becomes only Nglobal x Nglobal

If Nglobal 100 , the problem can be solved in seconds

kCkglobal Hk

HkT

k

=0

0

Cklocal 00

0

0 …

…

……

……

kwkxk

kwkk

… ………

Nglobal Nlocal x Ntraces

Matrix Inversion

Other Interesting Techniques

• Kalman Filter Alignment – CMS– Iterative– Updates alignment constants immediately

after each track

• SLD – Residuals as a function of misalignments– Fit residuals as a function of position– Determine alignment constant from matrix

inversion