Rivelatori a pixelsirad.pd.infn.it/scuola_legnaro_2007/Presentazioni_pdf/04_Bolla.pdf · 525...

Sep. 25th 2006 G. Bolla Purdue U. Vertex2006 1

Rivelatori a pixelIl rivelatore a pixel di CMS come un “case study”

G. [email protected]

Purdue University Department of Physics

525 Northwestern AvenueWest Lafayette, IN U.SA 47907-2036

26 marzo 2007 Legnaro, It 2

Outline

Motivazioni per rivelatori a pixel in esperimenti di fisica delle alte energieDescrizione dei rivelatori a pixel

SensoriElettronica di front-end

InterconnessioniDesscrizione dei moduliPerformance


Risultati ottenuti con rivelatori divertice a microstrisce al Tevatron

b-quarks decay with long lifetimeEssential to distinguish secondary vertexes from the primary beam-beam interaction to tag bs.High resolution position detector as close as possible to the interaction regionMicrostrip did (and are still doing) the job at the Tevatron

Top quark discovery in the 90s at Fermilab


The requirements for the Large Hadron Collider (LHC) at CERN

At LHC the physics motivations are similar but:

The track density is so high that is mandatory to reduce the cell size to unambiguously associate hits to traks. Single cell reduced in size by 2-3 orders of magnitude.from 10 cm x 50 μm strip to 150 μm x 100 μm pixels


Dalle microstrisce ai pixels


Cos’e un rivelatore a pixel?

Reverse biased Diode array (sensor) interconnected to the FE electronics (Read-Out-Chip) by flip chip technology (bumps) and controlled/read-out through I/O pads connected via wirebonds


Challenges a LHC/CMS

SpeedBeam-beam interaction every 25nsParticle rate as high as 20 MHz/cm2

Trigger latency 3.2 μsRadiation hardness

6E14 1MeV n equivalent in the first 4-6 years of operationsLimited maintenance4T magnetic field Precision

20-30 μm single hit resolution on both Φ and ZLow power consumptionLow cost


L’implementazione a CMS

~1 m0.3 m

man

The core of the CMS tracking system is a (hybrid) Pixel detector

100 × 150 µm2 pixel size ⇒ excellent spatial resolution ≈10-20 µm

Charge sharing promoted by 4T B field and 200 tilt in FPiX

3 barrel layers (BPix)r=4.3, 7.2 and 11 cm672 full modules and 96 half modules11528 ROCS, 48 MpixelsTotal area 0.78 m2

4 disks (FPix)Z=±34.5 and ±46.5 cm (∼ 6 cm above beam line)96 blades with 672 modules4320 ROCS, 18 MpixelsTotal area 0.28 m2

Operation at∼ -10 0C in 4 T B field3 space points up to |η| < 2.25


I sensori

Challlenges and solutionsHigh voltage operation: Multi-Guard-Ring structures at the sensor peripheryHomogeneous charge collection efficiency (CCE) on the whole pixel area: minimization of the interpixel isolation structures

n-on-n for partial depletion operation post bulk inversionP-stop interpixel isolationMultiguard peripheryDouble sided process (5+5 masks)Single wafer design

Activearea

Scribeline


I sensori

Barrel (p-spray) Disk (p-stops)

Within the CMS experiment two different choices have been made:Moderated p-spray for the barrel sensorsP-stop open rings for the disks sensors

NIM A 501 (2003) 160–163NIM A 511 (2003) 106–111NIM A 485 (2002) 89–99


Angolo di Lorentz

Signal charge is deflected by the magnetic field.The resolution is improved by the enhanced charge sharingNeed for sophisticated simulations and clustering algorithms

The effect is reduced by radiation damaged

• Lower charge collection depth• Drift velocity saturation due to the

large electric field applied


Disegno a turbina (dischi)

To optimize the advantages from the Lorentz angle the disks are configured with a turbine line geometry.20 deg rotation of the modules w.r.t. the plane of the diskAnt-symmetric disks w.r.t. the interaction point

+

-

B

B-plus-sidebps

B-minus-sidebms

+Z

+Y

+X+

-

B

B-plus-sidebps

B-minus-sidebms

+Z

+Y

+X


FE Read-Out-Chip (ROC)

0.25 μm IBM.Mixed Analog/Digital35 I/O padsLVDS signalsI2C controlOverall a little less than a square cm8+ years of development


Schematica del singolo pixel

Single pixel cell with:•Preamplifier•Shaper•Comparator•Discriminator•Pixel addresses (ROW)


Performance

Pulse heightLinear up to 2 m.i.p.s

Time walkLess than 20ns from 2000 to 20000 e-


Performance

Threshold

Before trimming σ ≈ 293 e- After trimming σ ≈ 80 e-

Threshold tunable to ~2000 e- with small dispersion


READOUTA pixel detects a hit and sends down to the column periphery the fast=OR

Latches the time stampInitiate the scan of the double column by the token

The token drains the data from the hit pixels in the double column into the buffers at the periphery.

All 26 double columns act independently and asynchronously from each other

No clocks in the active areaData in the buffers are discarded after the trigger latency unless a trigger is presentData from all double columns are serialized and sent out on a serial analog output with the decoded pixels addresses and the integrated charge


L’output analogico

Two bits for the column address and three for the row.Last piece of information is the PH (Pulse Height or integrated charge) Addresses are on a base 5 code

Minimize the readout-time


Data Losses


Data losses

High rate test of PSI 46 chip at PSI where a small sample was bombarded with pionsInefficiency = fraction of empty readouts < 3% at 80MHz/cm2Not easy to compare with the CMS environment


Le interconnessioni: bumpbonding

Indium or PbSn technologies


Pros and cons

IndiumNo need for high temperature processingPossible to do it in house (PSI and UCDavis)Low mechanical strength of the bondNot many vendors

PbSnMore of an “standard” technology (IZM, RTI)Expensive (at least at this point in time)


Metallurgia


Processo fotolitografico

Multiple steps photolithographic process.4” sensor wafers8” ROC wafersLimited number of vendors available


Flip chip

Components are heated to melting pointAligned and joined to each other


YieldDisconnected channels usually in the 10-4.Number of components damaged during the process in the few %.Still dedicated (and small) production lines.Normal pull strength: 1g/bond


The modules

Barrel module with 2x8 ROCs

Different geometries for the disks (plaquettes)


Modulo in sezione (disk)

Adhesive between the silicon substrate and the flex circuit.•Thermal contact for cooling

•Radiation damageCho-term 186 3M 9882

VHDI details


Mechanical support and cooling

•Here engineers really get creative

•Al Brazed cooling•Berilium panels

•LOW MASS


Disco


Componenti meccaniche per ilraffreddamento


Overview of the CMS Pixel Disks system

Detector will be installed with four ½ cylinders

½-Disks:12 Blades

Blade:2 Panels

TBM HDI,Plaquettes

Plaquette:Sensors,

ROC, VHDI

Pigtail:Panels to

Adapter BoardFan-in/Fan-out

Extension Cable

Port Card: AOH (2),DOH, ALT, Gate Keeper,

TPLL, DCU

FEC

FED

CMS DAQ

½-ServiceCylinder

O-fibers

Power,Cooling

Model

Note that the model on the left still has 3 disks while we are building only 2 on each side of the IP.


Performance: test beam

XY120 GeV π

O • •• • •••Y(3) Y(11)X(4) X(5) X(9) X(10)CMS Z

•Fermilab Test beam 120 GeV p•BTeV pixels telescope

•FPiX Reconstruction efficiency •(98.8 ± 0.1)% for plaquette irradiated up to φ =8× 1014Neq/cm2

ε=1


Performance: risoluzione

2 columns - 1 row events:Charge sharing in x direction

σx=10.2 μmσx=10.2 μm res(x)=9.0 μmres(x)=9.0 μm

1 column - 2 row events:Charge sharing in y direction

σy=9.0 μmσy=9.0 μm res(y)=6.7 μmres(y)=6.7 μm

Pitch / For single pixel hits greatly improves on 2 pixel hits12

Results are for normal incidence without any magnetic field

No data available with 20 deg tilt angle and/or B field


Performance: test beam

All triggers from scintillator

Overlap with telescope

Plaquette irradiated to 8E14Potential for microdischarges (pop-corn noise) at the pstop edges

No evidences of any appreciable noise occupancy.

Operations up to 700 V without any sign of breakdownParticle detection efficiency comparable with the pre-irradiation values (>99%)

LaserQ vs V

100V 400V 700V


Conclusions

I rivelatori a pixels erano l’unica possibilesoluzione per I rivelatori di vertice degliesperimenti a LHC durante la fase di sisegnoLa technologia sviluppata e’ affidabile e I rivelatori per ATLAS, CMS e Alice sono in avanzata fase di costruzione e presto verranno installati nei rispettivi experimentiLe prestazioni dei dispositivi sviluppatieccedono le domande di disegno.La stabilita’ delle prestazioni e’ stata misuratafino a fluenze integrate dell’ordine di 1E15.

Rivelatori a pixelsirad.pd.infn.it/scuola_legnaro_2007/Presentazioni_pdf/04_Bolla.pdf · 525...

Documents

Transcript of Rivelatori a pixelsirad.pd.infn.it/scuola_legnaro_2007/Presentazioni_pdf/04_Bolla.pdf · 525...