Semiconductors as particle detectorsheraeus-technology.desy.de/e8/e44905/Gregor... ·...

Preview:

Citation preview

Semiconductors as particle detectors

Ingrid-Maria Gregor, DESY

Thanksto:MarcWinter,LaciAndricek,CinziadaVia,PaulaCollins,UliKoetz,JimVirdee,CarstenNiebuhr,FrankSimon,FabianHügging,MarkusChrisDanzani,LutzFeld,RobertKlanner,ChristophedelaTaille,NorbertWermes

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Outline

MoDvaDonBasicsofSemiconductorDetectorsStripDetectorsPixelDetectorsPixelsforFutureExperimentsSummary

FieldofsemiconductordetectorsforparDcledetecDonisextremelylarge

Cangiveonlyaroughoverview

Myviewisbiased

alotofsilicon

morepixelsthanstrips

almostonlyHEP

Thefirsttransistor,inventedatBellLaboratories1947

2

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Motivation

Semiconductorshavebeenusedinpar?cleiden?fica?onformanyyears:

~1950:Discoverythatpn‐JuncDonscanbeusedtodetectparDcles.

Semiconductordetectorsusedforenergymeasurements(Germanium)

Since~30years:SemiconductordetectorsforpreciseposiDonmeasurements.

preciseposiDonmeasurementspossiblethroughfinesegmentaDon(10‐100μm)

mulDpliciDescanbekeptsmall(goal:<1%)

TechnologicaladvancementsinproducDontechnology:

developmentsformicroelectronics

ZEUSMVD2000

3

DELPHIVFT1996

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

CDFSVXIIa(2001‐)

~11m2siliconarea

~750000readoutchannels

DELPHI(1996)

~1.8m2siliconarea

~175000readoutchannels

Large Silicon Systems

CMSSiliconTracker(~2007)

~12,000modules

~223m2siliconarea

~25,000siliconwafers

~10Mreadoutchannels

4

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Basics of semiconductor Detectors

Ingr

id-M

aria

Gre

gor,

Sem

icon

duct

ors

as P

artic

le D

etec

tors

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Largegap:thesolidisaninsulator.

Nogap:itisaconductor.

Smallbandgap:semiconductor

Forsilicon,thebandgapis1.1eV,butittakes3.6eVtoionizeanatom.TherestoftheenergygoestophononexitaDons(heat).

Inagas,electronenergylevelsarediscrete.Inasolid,energylevelssplitandformanearly‐conDnuousband.

Semiconductor Basics I

6

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

n‐type:Inann‐typesemiconductor,negaDvechargecarriers(electrons)areobtainedbyaddingimpuriDesofdonorions(eg.Phosphorus(typeV))

DonorsintroduceenergylevelsclosetoconducDonbandthusalmostfullyionized=>FermiLevelnearCB

Electronsarethemajoritycarriers.

p‐type:Inap‐typesemiconductor,posiDvechargecarriers(holes)areobtainedbyaddingimpuriDesofacceptorions(eg.Boron(typeIII))

Acceptorsintroduceenergylevelsclosetovalencebandthus‘absorb’electronsfromVB,creaDngholes=>FermiLevelnearVB.

Holesarethemajoritycarriers.

Doping Silicon

7

p‐andn‐dotedsemiconductorcombined

GradientofelectronandholedensiDesresultsinadiffusemigraDonofmajoritycarriersacrossthejuncDon.

MigraDonleavesaregionofnetchargeofoppositesignoneachside,calledthedepleDonregion(depletedofchargecarriers).

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

PN-Junction

ArDficiallyincreasingthisdepletedregionbyapplyingareversedbiasvoltageallowchargecollecDonfromalargervolume

with

8

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Principle of semiconductor Detectors

1. CreaDonofelectricfieldvoltagetodepletethicknessd

:dopingconcentraDon

2. Keepdarkcurrentlow

:chargecarrierlifeDme

3. IonisingparDclescreatefreechargecarrier

4. Chargecarrierdriltoelectrodesandinducesignal

9

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Material Properties

Si Ge GaAs CdTe Diamant SiC

bandgap 1.12 0.67 1.42 1.56 5.48 2.99

energyfore‐ppair[eV] 3.6 2.9 4.2 4.7 13.1 6.9

e‐forMIP(300µm) 24000 50000 35000 35000 9300 19000

Z 14 32 31+33 48+52 6 14+6

Siliconistheonlymaterialwhichcanbeproducedinlargerwafersinhighquality

comparetokT=0.026eVatroomtemperature‐>darkcurrentundercontrol

highdensitycomparedtogases:ρ=2.33g/cm3

goodmechanicalstability‐>possibletoproducemechanicallystablelayers

largechargecarriermobility

fastchargecollecDonδt~10ns

Whyissiliconusedmoreolen?

10

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Protons in Silicon

0.4keV/µm‐>3.6eVcreateselectronholepair =>~110electron‐holepairsperµm(meanvalue)most probably number: 80 electrons

11

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Radiation Damage

ImpactofRadiaDononSilicon:

SiliconAtomscanbedisplacedfromtheirlatceposiDon

Pointdefects(EMRadiaDon)

Damageclusters(NuclearReacDons)

Importantinthiscontext:BulkEffects:Latcedamage:GeneraDonofvacanciesandintersDDalatoms(NIEL:NonIonizingEnergyLoss)

Surfaceeffects:GeneraDonofchargetraps(Oxides)(byionizingenergyloss)

Fillingofenergylevelsinthebandgap➭directexcitaDonnowpossible➭higherleakagecurrent➭morenoise➭“Chargetrapping”,causinglowerchargecollecDonefficiency

Canalsocontributetospacecharge:Higherbiasvoltagenecessary. 12

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Consequences of Radiation Damage

n

n+

p+

n

n+

p+

++ +

+ ++

--

--

--

++ +

+ ++

--

--

--

) )

))

Chargetrappingindefects

Macroscopicconstant:leakagecurrentanddepleDonvoltage

CountermeasuresGeometrical:developsensorsthatcanwithstandhigherdepleDonvoltages

Thinnersensors(butFEelectronicswithhighersensiDvityneeded)

Environment:sensorcooling(~‐10C)

Slowingdownof“reverseannealing”

Lowerleakagecurrents13

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Some Words onFront-End Electronics

Ingr

id-M

aria

Gre

gor,

Sem

icon

duct

ors

as P

artic

le D

etec

tors

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Overview of readout electronics

Most front-ends follow a similar architecture

Verysmallsignals(fC)‐>needamplificaDon

Measurementofamplitudeand/orDme(ADCs,discriminators,TDCs)

Severalthousandstomillionsofchannels

example:ALICEPixeldetector

15

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Very Large Scale Integration

VLSIenables

highchanneldensity

pre‐amplificaDon,datastorageetc.veryclosetothedetector

reducednoise

lowpowerdissipaDon

industrialproducDon

integraDondensityisgrowingrapidly

16

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Industry Scaling Roadmap

New generation every ~2 years with α = √2from 1970 (8 µm) to 2009 (35 nm) (industrial application)End of the road ? Power dissipation sets limits HEP nowadays at 90nm and 130nmProblem: by the time a technology is ready for HEP -> “old” in industry standards

HEP

17

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Radiation effects on CMOS: ionizing

Decreaseoffeaturesize:higherradiaDontolerance:

PosiDvechargetrappedingateandfieldoxidesTrappedchargedissipatesbytunnellinginthin‐oxidetransistors

RadiaDontolerantlayouttechniquesdesignedbyCERNRD49in0.25µmtoavoidparasiDctransistorleakage

Source

Drain Gate

Guard

Bird’s beak

Field oxide

Parasitic MOS

Parasitic channel

Standarddevicelayout

EnclosedlayoutTID on IBM 130nm NMOS [F. Faccio CERN]

18

gateenclosesalln+regionsavoidinganythicktransistorrelevantoxidestructures

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Strip Detectors

Ingr

id-M

aria

Gre

gor,

Sem

icon

duct

ors

as P

artic

le D

etec

tors

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Strip Detectors

FirstdetectordevicesusingthelithographiccapabiliDesofmicroelectronics

FirstSilicondetectors‐>stripdetectors

Canbefoundinallhighenergyphysicsexperimentsofthelast20years

Principal:Siliconstripdetector

p+-strips

metallization (Al)

depletion voltage

n-siliconionizingparticle

n+-siliconmetallization (Al)

80 e-h/µm

ArrangementofstripimplantsacDngaschargecollecDngelectrodes.

Placedonalowdopedfullydepletedsiliconwafertheseimplantsformaone‐dimensionalarrayofdiodes

ByconnecDngeachofthemetalizedstripstoachargesensiDveamplifieraposiDonsensiDvedetectorisbuilt.

TwodimensionalposiDonmeasurementscanbeachievedbyapplyinganaddiDonalstriplikedopingonthewaferbackside(doublesidedtechnology)

20

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

First HEP Application: NA11

Alerdiscoveryofcharm(1974),τ‐lepton(1975)andbeauty(1977)withlifeDmescτ~100μm:needfast(ns),andprecise(μm)electronictrackingdetectors

stripdetectorforNA11in1981

1200strip‐diodes

20μmpitch

60μmreadoutpitch

24x36mm2acDvearea~0.01m2

posiDonresoluDon~5.4μm

8layeratthestart

precisetrackreconstrucDon

readoutelectronic:~1m2!

21

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Si Microstrip Detectors for LHC

Early1990’s:Atthe?meoftheConceptual

DesignoftheppExperiments

RadiaDondamagepoorlyunderstood

Cost/unitareawasprohibiDvelylarge

Largeno.ofchannelsrequired

Whatwasknown:

leakagecurrentincreasedlinearlywithfluence

typeinversion–higherandhigherbiasvoltage

lrequired

reverseannealing

What was doneHVbehaviourimprovedbycarefulprocessinganduseofmulDpleguardringsSidetectorshadtobekeptpermanentlycoldFastpre‐amplifiersdevelopedtocopewith25nscollidingbunchesLeakagecurrentdealtwithfastamplifiersCost/unitareasignificantlyreducedbygrowinglargerdiameteringots(6”insteadof4”),single‐sidedprocessing(p‐on‐n)ImplementaDonoffront‐endread‐outchipinindustrystandarddeepsub‐microntechnology

22

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

CMS Si-Tracker

Si‐Strip‐Detector:

~210m2Silicon

25000Sensors,9.6Mchannels

10barrellayers,2x9discs

Thelargesteverbuiltsilicontracker

23

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

CMS Si Barrel

Single and double sided layers

CMSInnerbarrelSiTracker:Single‐SidedSi‐Strip

24

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

ATLAS SCT

ATLAS Si-Detector SCT: Si- strips: 4 Barrel-layer, 2 x 9 discs

25

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

ATLAS SCT

SCT strips: 61 m2 silicon, ~6.2 M channels4088 modules, 2112 barrel (1 type), 1976 in the discs (4 different types)

26

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

ATLAS SCT Module

barrel-module

27

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

ATLAS SCT Module

barrel-module

disc-module27

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Pixel Detectors

Ingr

id-M

aria

Gre

gor,

Sem

icon

duct

ors

as P

artic

le D

etec

tors

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Limits of Strip Detectors

In case of high particle fluences ambiguities give difficulties for the track reconstruction

DerivingthepointresoluDonfromjustonecoordinateisnotenoughinformaDontoreconstructasecondaryvertex

PixeldetectorsallowtrackreconstrucDonathighparDcleratewithoutambiguiDes

GoodresoluDonwithtwocoordinates(dependingonpixelsizeandchargesharingbetweenpixels)

‣ Veryhighchannelnumber:complexread‐out

‣ ReadoutinacDveareaadetectorFirstpixels(CCDs)inNA11/NA32:~1983

29

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Hybrid Pixels – “classical” Choice HEP

chip

sensor

Theread‐outchipismounteddirectlyontopofthepixels(bump‐bonding)

Eachpixelhasitsownread‐outamplifier

Canchooseproperprocessforsensorandread‐outseparately

Fastread‐outandradiaDon‐tolerant

…but:

Pixelareadefinedbythesizeoftheread‐outchip

HighmaterialbudgetandhighpowerdissipaDon

HybridPixel(CMS)

CMSPixels:~65Mchannels150µmx150µm

ATLASPixels:~80Mchannels50µmx400µm(longinzorr)

Alice:50µmx425µmLHCb

Phenixupgrade

Fair

CBM

PANDA 30

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Apixelmodulecontains:1sensor(2x6cm)~40000pixels(50x500mm)16frontend(FE)chips2x8arraybumpbondedtosensorFlex‐hybrid1modulecontrolchip(MCC)Thereare~1700modules

ATLAS-Pixels

31

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

ATLAS-Pixels

32

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Services!!

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Services!!

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Material Budget of LHC Experiments

Atlas

CMS&AtlasbothslippedconsiderableinkeepingX/X0originallyaimedfor!

OldargumentthatSiliconwouldbetoothickisnotreallytrue==>power&cooling

CMS

34

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Next Generation Pixel Detectors

Ingr

id-M

aria

Gre

gor,

Sem

icon

duct

ors

as P

artic

le D

etec

tors

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Current Challenges (ILC, sLHC)

Mainchallenge:idenDfycquarkandleptonjets

lifeDme~10‐12sec=>~100um

=>parDclesdecaywithinthevacuumbeampipe

reconstructdecayproducts

Alsohere:Trendintrackingdetectors:pixelliseddetectorsinstalledveryclosetothebeaminteracDonregion

MinimaldistancelimitaDons:

• beampiperadius

• beamassociatedbackgrounds

• densityofparDclesproducedattheIP

Consequencesonoccupancyandradia?onlevel

36

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Optimising = Compromising

ConflictbetweenphysicsperformancedrivenparametersandrunningcondiDonconstraints:

Physicsperformance:spaDalresoluDonandmaterialbudget(+distancetoIR)

RunningcondiDons:read‐outspeedandradiaDontolerance

Moreover:

➡ limitaDonsfrommaximumpowerdissipaDoncompaDblewithrunningcondiDonsandmaterialbudget

➡ limitaDonsfromhighestdataflowacceptablebyDAQ

UlDmateperformanceonallspecificaDonscannotbereachedsimultaneously

eachfacility&experimentrequiresdedicatedopDmizaDon(hierarchybetweenphysicsrequirementsandrunningconstraints

thereisnosingletechnologybestsuitedtoallapplicaDons

explorevarioustechnologicalopDons

moDvaDonforconDnuousR&D(opDmumisstronglyDmedependent)

37

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

physics or Running Conditions

Physicsperformancedriven:

thin(potenDallyundepleted)sensiDvevolume

ILC,RHIC,CLIC,SuperB,FAIR

CMOSsensors,CCDs,DEPFETs,VerDcallyintegrated(“3D”)

Runningcondi?onsdriven:

”thick”depletedsensiDvevolume

LHC&SLHC

Hybridpixelsensors,3Dsensors,VerDcallyintegrated(“3D”)

Future : 3D integrated pixel devices reduce the gap between the two main

optimization optionsTwotypesof3D:3Dsensors‐>CinziaDaVia3Dintergrateddevices‐>ChrisDanKiesling

38

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Harsh Running Conditions - SLHC

Improvement/evoluDonofhybridpixelsensors:SmallerCMOSfeaturesizemorecompactFEμcircuits

smallerpixelsoccupancy

ImprovedsensiDvevolumeradiaDonhardness

LargernumberofpixelspowerdissipaDonisanissue!

AlternaDvestohybridpixels:ParDcularlyinfashion:3Dsensors

Others:3Dintegrateddevices

300–400pile‐upeventsatstartoffill

wanttosurviveatleast3000�‐1datataking

B‐layerat37mm:

~30trackspercm‐2perbunchcrossing

>10161MEVn‐equivalentnon‐ionising

Few10sofMGray(10xLHC)

39

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

ThinplanarpixelsfortheinnerlayersofthenewATLASpixelsystematSLHC:Atthesamevoltagethin(=overdepleted)detectorsareexpectedtohaveahigherelectricfieldthanthick(=parDallydepleted)detectors.Theycontributetokeepthematerialbudgetlow,whichisextremelyimportantintheinnerlayerstomaintaingoodtrackingperformances.

Results of a first characterization of the pixel structures on p-type wafers:

Lowleakagecurrents(<7nA/cm2for75μmstructures,<15nA/cm2for150μmstructures).DepleDonvoltagesaround20and80Vforthe75μmand150μmthickstructuresrespecDvelyHighVbreakforallthepixelstructuresproduced

Thin Planar Pixel Sensors

40

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

3D Sensors

3‐darrayofpandnelectrodesthatpenetrateintothedetectorbulkLateraldepleDon:

Max.drilanddepleDondistancesetbyelectrodespacingReducedcollecDonDmeanddepleDonvoltageThickerdetectorspossibleLowchargesharing

BUT:non‐standard(planar)technology

Bothelectrodetypesareprocessedinsidethedetectorbulkinsteadofbeingimplantedonthewafer'ssurface.

41

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

The Vertex Detector at the ILC

Need: Goodangularcoveragewithmanylayersclosetovertex:

|cosθ|<0.97.

Firstmeasurementatr~15‐16mm.5‐6layersouttor~60mm.

EfficientdetectorforverygoodimpactparameterresoluDon

Material~0.1%X0perlayer.

CapabletocopewiththeILCbeamstrahlungsbackground

ModestaveragepowerconsumpDon<100WSinglepointresoluDonbe�erthan3μm.

smallpixels,thinsensors,thinr/oelectronics,lowpower(gascooling)

Measureimpactparameter,chargeforeverychargedtrackinjets,andvertexmass.

FigureofmeritfortheVXD:

ImpactParameterResoluDon

(MarcWinter)

Accelerator a(µm) b(µm)

LEP 25 70

SLD 8 33

LHC 12 70

RHIC‐II 13 19

ILC <5 <10

42

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

The Vertex Detector at the ILC

Innerlayer1.6MPixelsensorsOnceperbunch=300nsperframe:toofast

Oncepertrain~100hits/mm2:tooslow

5hits/mm2=>50μsperframe:maybetolerable (Note:fastestcommercialimaging~1ms/MPixel)

FourdifferenttechnologiesunderstudyforILCvertexdetectorCCD,DEPFET,CMOS,and3DdifferentvariantsofeachtechnologyapproachunderinvesDgaDon

Themainsplitisbetweenthese2categories:conDnuousread‐outinsidetrainsdelayedread‐outinbetweenconsecuDvetrains

369 ns

x2625

0.2 s

~1 ms

BunchTrain

BunchSpacing

43

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

ILC: What's on the Pixel Market?

Accumula?onof~70‐150BX

1.cont.r/oduringtrain

CPCCD(LCFI)DEPFETMimosa(Strasbourgatal.)

2.storeandr/oinpause

CAPs(Hawaii)ISIS(LCFI)

Accumula?onofabout3000BX

FP‐CCD(KEK,JAXA/ISAS,TohokuUni)

single‐bunch?mestamping"hybridpixelsw/obumps"

3Dintegratedpixels(Fermilab)

SOI(Fermilab,LBNL)

ChronoPixels(Yale/Oregon)

DeepN‐WellMAPSSDR(INFNMilan,Pavia,RomaIIIUni.Bergamo,Insubria,Pavia)

Currentlythereareabout~10“candidates”fortheILCVTXDetector.

ThesetechnologieshavedifferentapproachestocopewithbeaminducedbackgroundattheILC

Allapproachesaimfor~3µmprecisionand<40mm2‐hitresoluDon

Targetmaterialbudgetis~0.1%X0perlayer

44

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

DEPFET

DePletedFeld Effect Transistor fullydepletedsensiDvevolume

fastsignalriseDme(~ns),smallclustersize

nochargetransferneeded

fasterreadout

be�erradiaDontolerance

internalamplificaDon

largesignal,evenforthindevices

r/ocap.independentofsensorthickness

charge‐to‐currentconversion:gq=dId/dq≈1nA/electron,scaleswithgatelength

ChargecollecDonin"off"state,readoutondemand

Rowwiseread‐out("rollingshu�er”):selectrowwithexternalgate,readcurrent,clearDEPFET,readcurrentagainthedifferenceisthesignal

onlyonerowacDvelowpowerconsumpDon

But:for40kHzframerate25nsperrowlDEPFETisbaselinetechnologyforthevertexdetectorofSuperBelle

45

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Monolithic Active Pixels (MAPS)

e.g.Mimosa(MinimumIonizingParDcleMOSAcDvePixelSensor)

h

AcDveareaunderneaththeelectronics(epi‐layer<20µmthick)providing100%fill‐factorChargegeneratedbyionizaDonintheepitaxiallayerthermallydiffusetowardlowpotenDaln‐wellregionStandard,cost‐effecDveCMOSprocess,nopost‐processing

<20µm

FeaturesoftheMIMOSA–detectors:

SinglepointresoluDon1µm–3µm

Pixel–pitch10‐40µm

Thinningachieved50‐120µm

S/NforMIPs20–40

DetecDonefficiency>99%

RadiaDonhardness:1MRad;2x1013neq/cm²

ProducedinvariouscommercialCMOS‐processes

MimosaisbaselinetechnologyforthevertexdetectorofSTAR

46

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

CCD based

Charge‐CoupledDevices(CCDs)Demonstratedinlargesystem(307Mpx)atSLD,butslow.

ColumnParallelCCDs(CPCCD)columnparallelreadout,withbump‐bondconnecDonson20µmpitchtoreadoutchipincludingamp,analogueCDS,ADCs,sparsificaDonandmemory

FPCCD(FinePixelCCD)–fullydepletedepi‐layertosuppressdiffusion

with5µmpixels,readoutoncepertrain;20Dmesfinerpixelgranularityinsteadof20Dmeslices

Image Sensor with In-Situ Storage (ISIS)Combines CCDs, active pixel transistors and edge electronics in one deviceCharge collected under a photo-gateCharge is transferred to 20-pixel storage CCD in situ, 20 times during the ~ 1 ms long bunch train

47

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

DoubleCMOSPixelMacro(50μmpitch)forDmingMicro(5μmpitch)forpreciseposiDonBufferdataduring~3000bunchesinatrainandreadoutbetweenbunchtrains

SOIdetectorsareafirststeptoward3DintegraDonsinceitusesmanyofthesameprocessesas3DintegraDon(oxidebonding,waferthinning,viaformaDon)

ThintoplayerwithsiliconislandsinwhichPMOSandNMOStransistorsarebuilt.Aburiedoxidelayer(BOX)whichseparatesthetoplayerfromthesubstrate.HighresisDvitysubstratewhichformsthedetectorvolume.DiodeimplantsareformedbeneaththeBOXandconnectedbyvias.

Chronopixel and SOI

48

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Silicon detector size 1981 - 2006[m

2 ]

Space

HEP

49

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Summary

Solidstatedetectorsplayacentralroleinmodernhighenergyandphotonphysics

UsedintrackingdetectorsforposiDonandmomentummeasurementsofchargedparDclesandforreconstrucDonofverDces(speciallypixeldetectors)

Byfarthemostimportantsemiconductor:Silicon,indirectbandgap1.1eV,however:3.6eVnecessarytoformehpair

AdvantagesSi:largeyieldingeneratedchargecarriers,finesegmentaDon,radiaDontolerant,mechanicallystable,…

Workingprinciple(general)diodeinreversebias(pnjuncDon)

Important:S/Nhastobegood.Noise~1/Cforsystemsthatmeasuresignalchargesmallerfeaturesizesaregood.Pixel!

PixeldetectorsareusedinmostmajorcurrentparDcledetectorsandareplannedforfutureexperiment

R&Dforsemiconductordetectorsalwayshastobeontheedgeoftechnology

50

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Noise

detectorcapacity

darkcurrent

serialresistor

SegmentaDonintomanychannels

Noise Sources

51

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Resolution of Tracking Detectors

DependingondetectorgeometryandchargecollecDon

Strippitch

Chargesharingbetweenstrips

Simple case: all charge is collected by one stripTraversing particle creates signal in hit stripFlat distribution along strip pitch; no area is pronounced

Probability distribution for particle passage:

Simplecase:allchargeiscollectedinonestrip

Thereconstructedpointisalwaysthemiddleofthestrip:

52

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Resolution of Tracking Detectors II

CalculaDngtheresoluDonorthogonaltothestrip:

ResulDnginageneralterm(alsovalidforwirechambers):

Forasiliconstripdetectorwithastrippitchof80µmthisresultsinaminimalresoluDonof~23µm

Incaseofchargesharingbetweenthestrip(signalsizedecreasingwithdistancetohitposiDon)

ResoluDonimprovedbycenterofgravitycalculaDon

53

54

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Interactions of Particles with Matter

MaximumkineDcenergywhichcanbetransferredtotheelectroninasinglecollision

ExcitaDonenergy

DensitytermduetopolarizaDon:leadstosaturaDonathigherenergies

ShellcorrecDonterm,onlyrelevantatlowerenergies

54

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Interactions of Particles with Matter

MaximumkineDcenergywhichcanbetransferredtotheelectroninasinglecollision

ExcitaDonenergy

DensitytermduetopolarizaDon:leadstosaturaDonathigherenergies

ShellcorrecDonterm,onlyrelevantatlowerenergies

“relaDvisDcrise”“minimumionizingparDcles”βγ ≈ 3-4

“kinemaDcterm”

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Landau Distribution Silicon

mostprobableenergyloss

meanenergyloss

Bethe‐Blochdescribesaverageenergyloss

CollisionsstochasDcnature,henceenergylossisdistribuDoninsteadofnumber.

FirstcalculatedforthinlayerswasLandau.HenceenergylossisLandaudistributed.

SignalproporDonaltoenergyloss

Example:MonolithicacDvepixelsensorsensiDvelayer14um

55

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Literature

SemiconductorDetectorSystems,

HelmuthSpieler,OxfordUniversityPress2005

PixelDetectors–FromFundamentalstoApplica?onsL.Rossi,P.Fischer,T.Rohde,N.Wermes,SpringerVerlag2006

Evolu?onofSiliconSensorTechnologyinPar?clePhysicsFrankHartmann,SpringerVerlag2009

56

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rsPlanar process

N-type siliconSiO2

BB

As

n+

P+

n‐typewafersareoxidizedat1030oCtohavethewholesurfacepassivated.

Usingphotolithographicandetchingtechniques,windowsarecreatedintheoxidetoenableionimplantaDon.Differentgeometriesofpadsandstripscanbeachievedusingappropriatemasks.

ThenextstepisthedopingofsiliconbyionimplantaDon.DopantionsareproducedfromagaseoussourcebyionisaDonusinghighvoltage.Theionsareacceleratedinanalectricfieldtoenergyintherangeof10keV‐100keVandthentheionbeamisdirectedtothewondowsintheoxide.P+stripsareimplantedwithboron,whilephosphorousorarsenicareusedforthen+contacts.

Anannealingprocessat600oCallowsparDalrecoveryofthelatcefromthedamagecausedbyirradiaDon.

ThenextstepisthemetallisaDonwithaluminium,requiredtomakeelectricalcontacttothesilicon.Thedesiredpa�erncanbeachievedusingappropriatemasks.

Al

ThelaststepbeforecutngisthepassivaDon,whichhelpstomaintainlowleakagecurrentsandprotectsthejuncDonregionfrommechanicalandambientdegradaDon.

57

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Sincethemicro‐stripdetectorprovidesonlyonecoordinatewithgoodprecision,thesegmentaDonofthebackplaneisanaturalwaytoprovideasecondcoordinateandthusaspacepointwithoutaddingmaterialonthetrajectoryoftheparDcles.

Theuseofdouble‐sidedmicro‐stripdetectorsallowsthecorrelaDonofsignalscollectedonthetwosides,whichapartfromthereadoutelectronicsnoiseandresponseisthesame,thusreducingmulD‐hitambiguiDes.

+ + + + + +- - - - - -

SiO2 Al

n+

Al

n+

Fixed oxide charge

R≈ few kΩ

n-type

R > few kΩSiO2

Subdividingsimplythen+contactsthepresenceofposiDvechargeattheSi‐SiO2interfaceinducesinthen‐typesubstrateanaccumulaDonlayerofelectrons,resulDnginalowresistancebetweenthestrips.Thereforethesignalspreadsovermanyelectrodes,makingthesubdivisionineffecDve.

backplane

backplane

Al Al Al

n+ n+p+n-type

+ + + + + + -

- - -

- -

Amethodusedtosolvethisproblemistoimplantap+blockingstripinbetweenthen+ones.TheblockingstripsarelelfloaDng,sincetheirfuncDonisjusttointerrupttheconducDonchannel.

Double-Sided Strip Detectors

58

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Tevatron: D0

Barrelanddisc:stripdetectors

840kchannels

OperaDonalsince2002

59

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

successfullylaunched11.June200816towermodules

37×37cm2acDvearea

2mminter‐towerseparaDontominimizetheinacDvearea

70m2ofSi(inspace!!!)

11500SSD8.95X8.95cm2,

384strips‐880,000channels

440μmthick

228μmstrippitch

GLAST Mission

60

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

diamond sensor

Polycrystallineandsinglecrystal

Lowleakagecurrent,lownoise

Lowcapacitance

RadiaDonhardmaterial

OperaDonatroomtemperaturepossible

Drawback:50%signalcomparedtosiliconforsameX0Butbe�erS/NraDo(nodarkcurrent)

growth

substrate

Grain size: ~100-150μm

SuccessfultestofscCVDdiamondpixelmodule

StableoperaDon

FullchargecollecDonat0.25V/µm

Goodefficiency: ε>99.9%

MeasuredResoluDon:σ=8.9µm(200V,normalincidence)

scCVDmoduleretains~80%ofiniDalchargecollectedalerirradiaDonto0.7x1015p/cm2

61

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Tevatron: CDF

Si‐stripdetector

specialfeature:Layer00directlyonbeampipe

720kchannels

OperaDonalsince2002

Inner Strip Layers (ISL):

62

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Bonding

63

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

Large scale silicon systems

The most critical parts are the sensors, ASICs and system engineering (mechanics, power, cooling, assembly, etc)

To develop and buy silicon sensors for several hundreds of m2 silicon sensors is not an easy task:

Extend previous Multi-Geometry studies to substrate thickness less than or equal the pitch

Strip/Pixel capacitance (back-plane, inter-strip/pixel & total)Critical fields, depletion and break-down voltageSensor functionality (charge collection efficiency etc)Detailed design parameters for masks

Extend previous studies from LHC to SLHC fluence – large irradiation programs neededExtend previous studies to include n-on-pRe-produce complementary sets of measurements and simulationStudy biasing, guard rings, isolation methods

64

65

100In

grid

-Mar

ia G

rego

r, Se

mic

ondu

ctor

s as

Par

ticle

Det

ecto

rs

The International Linear Collider

Parameters:

√s=500GeV,tunablefrom200to500GeV,upgradeableto1TeV

∫Ldt=500�‐1in4years(peakluminosity2∙1034cm‐2s‐1)

e‐‐e+collider:two11kmSClinacsat31.5MV/m

DualtunnelconfiguraDon(safetyandaccessibility)

SingleIR,crossingangle14mrad,twodetectorsinpush‐pulloperaDon

SC Nb Cavity