LHC Symposium - May 3, 2003 1 Super LHC - SLHC LHC Detector Upgrade Dan Green Fermilab

LHC Symposium - May 3, 2003 1

Super LHC - SLHCSuper LHC - SLHC

LHC Detector Upgrade

Dan Green

Fermilab


OutlineOutlinePhysics Basics

Z’ vsRapidity RangeMinbiasPileup and Jets

Occupancy and Radiation Dose Tracker UpgradeCalorimetryMuonsTrigger and DAQ

, sCERN-TH/2002-078 “Physics Potential and Experimental Challenges of the LHC Luminosity Upgrade”

10x will be challenging!


Mass “Reach” and LMass “Reach” and LThe number of Z’ detected in leptonic decays is:

For , if N = 100 is discovery level then M ~ 5.3 TeV is ~ the mass “reach” in 1 year (M=4 -> 5.3 TeV).

The leptons will be sharply limited to low |y| or large angles (“barrel”).

2

/

2{ [ } ( )( ) [ /8 ]( )]x M sW xu x xu x MN B e e y

11[ ( ) ( )] 0.36 (1 )xu x xu x x x


Mass Reach vs LMass Reach vs L

1032

1033

1034

1035

103

104

Luminosity(/cm2sec)

MZ'

(Ge

V)

N=100 Events, Z' Coupling

2 TeV 14 TeV 28 TeV 100 TeV

In general mass reach is increased by ~ 1.5 TeV for Z’, heavy SUSY squarks or gluinos or extra dimension mass scales. A ~ 20% measurement of the HHH coupling is possible for Higgs masses < 200 GeV. However, to realize these improvements we need to maintain the capabilities of the LHC detectors.

VLHC

LHC

Tevatron


KinematicsKinematics

Heavy States decay at wide angles. For example Z’ of 1 and 5 TeV decaying into light pairs. Therefore, for these states we will concentrate on wide angle detectors.

1 TeV5 TeV

/d dy

barrel y barrel


Inclusive InteractionsInclusive InteractionsThe inclusive p-p interaction has an inelastic, non-diffractive cross section ~ 50 mb.

It produces ~ equal numbers of which are distributed ~ uniform in rapidity, y, with a “density” ~ 9 pions per unit of y.

The pions have a distribution in transverse momentum with a mean, ~ 0.6 GeV.

I

(1/ ) /I d dy

Tp

, ,o


Detector EnvironmentDetector Environment

LHC SLHC

s 14 TeV 14 TeVL 1034 1035

100 1000

Bunch spacing dt 25 ns 12.5 ns

N( interactions/x-ing) ~ 12 ~ 62 dNch/d per x-ing ~ 75 ~ 375

Tracker occupancy 1 5Pile-up noise 1 ~2.2Dose central region 1 10

Bunch spacing reduced 2x. Interactions/crossing increased 5 x. Pileup noise increased by 2.2x if crossings are time resolvable.

2/( sec)cm 2/( sec)cm 1 /fb yr 1 /fb yr

Ldt


Pileup and LuminosityPileup and LuminosityFor ~ 50 mb, and = 6 charged pions/unit of y with a luminosity

and a crossing time of 12.5 nsec :

In a cone of radius = 0.5 there are ~ 70 pions, or ~ 42 GeV of transverse momentum per crossing. This makes low Et jet triggering and reconstruction difficult.

Ic

35 210 / seccm

95 10 int/ sec

62 int/

375 / ,

x

x ing

x ing unit of y


Z’(120) at L/5 and LZ’(120) at L/5 and L

020

4060

80

0

10

20

30

400

5

10

15

20

-7 -6 -5 -4 -3 -2 -1 00

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8Jet, Cone Rc = 0.8

log(x)

dr

-7 -6 -5 -4 -3 -2 -1 00

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8Jet, Cone Rc = 0.8, Centroid Offset in by /2

log(x)

drdR

Log(z), z = k/ET

ET(GeV)

Jet cone and 90 degrees to cone in


Z’(120) Mass ResolutionZ’(120) Mass Resolution

0 50 100 150 200 2500

5

10

15

20

25

30

35

40Z'(120) at Low Luminosity

Mjj(GeV)

0 50 100 150 200 2500

5

10

15

20

25

30

35

40

45

50Z'(120) at Design Luminosity

Mjj(GeV)

Note that the calorimeter cells are still fairly sparsely populated (granularity

) at 1034 . With the cuts shown, the dM/M with Gaussian fits is the same at L/5 and at L. Use the fact that QCD implies that there is a core of the jet at small dR and large z. Extend to 10x L using tracker and energy flow inside the jet? If x-ing is time resolvable, pileup is “only” 5x. Tracker can be used (energy flow) to remove charged energy deposits from vertices within the x-ing which are not of interest.

2(0.087)

M(GeV)


Tracker and Energy FlowTracker and Energy FlowFor 120 GeV Z’ match tracks in and to “hadronic” clusters within the jet. Improves dijet mass resolution. Units are HCAL tower sizes. Also use track match to remove charged pion deposits from pileup vertices ?

0 10 20 30 40 50 60 70 80 90-2

-1.5

-1

-0.5

0

0.5

1

1.5

2Z(120) Data, Match of Tracks to Calorimeter Clusters in

Et(GeV)

d

0 10 20 30 40 50 60 70 80 90-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1Z Data, Match of Tracks to Calorimeter Clusters in

Et(GeV)

dd

d

ET(GeV)


WW Fusion and “Tag Jets”WW Fusion and “Tag Jets”These jets have

~ pileup R = 0.5 and <y> ~ 3. Lose 5x in fake rejection. We must use the energy flow inside a jet cone to further reduce the fake jets due to pileup (~ uniform in R).

~ / 2T WE M

WW fusion

Pileup, R=0.5, |y|=3


Tracking DetectorsTracking Detectors Clearly, the tracker is crucial for much of the LHC physics [e.g. e, , jets (pileup, E flow), b tags].

The existing trackers will not be capable of utilizing the increased luminosity as they will be near the end of their useful life.

It is necessary to completely rebuild the LHC tracking detectors.


Tracker - OccupancyTracker - OccupancyThe occupancy, O, for a detector of area dA and sensitive time time dt at (r,z) is

e.g. Si strip 10 cm x 100 m in a 12.5 nsec crossing at r = 20 cm is 1.5 %For higher luminosity, decrease dA, or decrease dt (limit is x-ing time) or increase r – smaller, faster or further away.

2( ) /[2 ]I c dAdtO r


Tracker OccupancyTracker OccupancyPreserve the performance using :

Push Si strips out to ~ 60 cm. – developmentPush pixels out to 20 cm. – developmentFor r < 20 cm. Need new technologies – basic research

Shrink dA 5x at fixed r to preserve b tagging? If 12.5 nsec bunch x-ing, need 5x pixel size reduction.Possibilities

3-d detectors – electrodes in bulk columnsDiamond (RD42) - radhardCryogenic (RD39) – fast, radhardMonolithic – reduced source capacity.

21/ r


Monolithic Pixel - DEPFETMonolithic Pixel - DEPFET

Combine the detector and the readout for pixels?


Tracker – Ionizing DoseTracker – Ionizing DoseThe ionizing dose due to charged particles is:

The dose depends only on luminosity, r, and exposure time .

For example, at r = 20 cm, the dose is ~3 Mrad/yr – ignoring “loopers”, interactions, …. “naïve” expectation.

2[ / ( ' )] /[2 ]I c mipID dE d x r


Tracker ID vs. RadiusTracker ID vs. Radius

100

101

102

103

10-1

100

101

102

103 Ionizing Dose in Tracker for 10 35 L and 1 Year

r(cm)

Dos

e(M

rad)

naive

1 2 3

Define 3 regions. With 10x increase in L, need a ~ 3x change in radius to preserve an existing technology.


Tracking R&D -ITracking R&D -IRegion 1: r < 20cm

Occupancy -> Need pixels of a size factor ~ 5 smaller than used today (125x125 m2 -> ~ 50 x ~ 50 m2) -> benefit b-tagging

R&D: Pixels Sensor Technologies• new sensor materials – defect engineered Si, CVD diamond, SiC, passivated

amorphous Si etc.• 3-D detectors and new biasing schemes• Cryogenic Si tracker development• monolithic pixel detectors

Region 2: 20<r<60 cmNeed cell sizes 10 times larger than current pixels but at 10 times lower

cost/channel than current Si microstrips -> benefit p-resolution and pattern recognition

Si Macro-pixels of an area < 1mm2 : pads or shorter strips ?Could be upgrades of innermost Si strip layers of current detectorsR&D: to demonstrate low-cost macro-pixels concept, thin Si detectors.


Tracking R&D - IITracking R&D - IIRegion 3: r > 60 cm

Si-strips –decrease size of strips i.e. increase no. of channels by > 50%

Use standard ‘radiation resistant’ strip technology

R&D: Feasibility of processing detectors on 8” or 12’ Si wafers. Monitor commercial production progress.

Engineering

R&D: new materials, light weight, stable structures, cooling, alignment, implications for cryogenic operation, installation and maintenance aspects

Activation: 250 mSv/h – implications for access and maintenance

Cost: Reduce cost/channel by a factor of 10

Timescale : Need ~ 8-10 years from launch of R&D

~ 4 years to build, after ~ 4 years of R&D and prototyping ?


Electronics – Moore’s LawElectronics – Moore’s LawMicro-electronics: line-widths decrease by a factor 2 every 5 years. DSM (0.25 m) is radiation hard.Today 0.13 m is commercially available. In the lab 0.04 m, e.g. extreme UV lithography, is in existence. Expect trend will continue for a decade.

R&D Characterize emerging technologies

more radiation tolerance required – dose and Single Event Effects

advanced high bandwidth data link technologies

system issues addressed from the start

P. Sharp

Industry

Research

1m

10m

0.1m20001985


ECAL – Shower DoseECAL – Shower DoseThe dose in ECAL is ~ due to photon showers and is:

In the barrel, SD is ~ . In the endcap, SD ~

At r = 1.2 m, for Pb with Ec = 7.4 MeV, the dose at y=0 is 3.3 Mrad/yr, at |y|=1.5 it is 7.8 Mrad/yr.

2sin[ / ( ' )] [

s

/ ] /[2 ]

( / 2)[ / ]in

I o mip T c

T c

SD dE d x p E

ID p

r

E

2/[ sin ]r 32 2 3/[ ] ~ ( / )z z e


HCAL and ECAL DoseHCAL and ECAL Dose

0 1 2 3 4 510

-2

10-1

100

101

102

103 Dose in ECAL and HCAL for L = 1035 and One Year

Dos

e(M

rad)

The dose ratio is ~ . Barrel doses are not a problem. For the endcaps a technology change may be needed for 2 < |y| < 3 for the CMS HCAL. Switch to quartz fiber as in HF?

( ) /th cE p p E

naive

ecal

hcal


ECALECALFor both ATLAS and CMS the barrel will probably tolerate the increased dose. There are issues of ~ 2.2x increased pileup noise and poorer isolation for electrons. Shorter shaping times to resolve x-ing?

ATLAS LA has space charge and current draw issues. CMS has APD leakage current noise issues in the barrel. The CMS endcap needs development.


HCAL - CMSHCAL - CMSBoth ATLAS and CMS will function in the barrel region.

In the 3<|y|<5 region, a reduction to y < 4.2 keeps the dose constant. The loss of efficiency is not terrible (peak “tag”rate at |y|=3). Or replace quartz fibers with high pressure gas? Better tower granularity might be needed due to pileup and “fake” jets.

At |y| ~ 3 the CMS scintillator needs development – improved scintillator or go to quartz fibers ( volume degraded is quite small).


HCAL - CoverageHCAL - Coverage

Reduced forward coverage to compensate for 10x L is not too damaging to “tag jet” efficiency


Scintillator - Dose/DamageScintillator - Dose/Damage

Scintillator under irradiation forms Color centers which reduce the Collected light output (transmission loss). LY ~ exp[-D/Do], Do ~ 4 Mrad

This technology will not survive gracefully at |y| ~ 3. Use the technology that works at LHC up to |y|~ 5, quartz fibers?

|y|=2, 1 yr.


Muons and ShieldingMuons and ShieldingThere is factor ~ 5 in headroom at design L. With added shielding, dose rates can be kept constant if angular coverage goes from |y|<2.4 to |y|<2.

2/( sec)n cm

r

r

z


Trigger and DAQTrigger and DAQAssuming LHC initial program is successful, raise the trigger thresholds.Rebuild trigger system to run at 80 MHz. Utilize those detectors which are fast enough to give a BCID within 12.5 nsec (e.g. Calorimetry, Tracking).Examine algorithms to alleviate degraded e isolation, for example.Design for the increased event size (pileup) with reduced L1 rate and/or data compression.For DAQ track the evolution of communication technologies, e.g. 10 Gb/sec Ethernet.


300 GeV Pion – H2 test Beam300 GeV Pion – H2 test Beam

HTR - Bunch crossing number (LHC)

The shape of the pulse in time is ~ as expected – due largely to scint flours. Bunch crossing ID can be extended to 12.5 nsec ( 80 MHz) as established in test beam.

E


SummarySummaryThe LHC Physics reach will be substantially increased by higher luminosity.To realize that improvement, the LHC detectors must preserve performance.The trackers must be rebuilt – with new technology at r < 20 cm.The calorimeters, muon systems, triggers and DAQ will need development.The upgrades are likely to take ~ (6-10) years. Accelerator is ready ~ (2012, 2014). The time to start is now, and the people to do the job are those who did it for the present detectors - integration. However, new people are needed.


SLHC Detector - SummarySLHC Detector - SummaryTracking and b-tagging

Isolated high pT (> 20 GeV) tracks - it should be possible to maintain similar efficiency and momentum resolution without a tracker upgrade, for fixed b-tagging efficiency, rejection against light quarks will deteriorate by factor ~8 (pT ~ 50 GeV)

Electron identification and measurement For electron efficiency of 80% jet rejection decreases by ~ 50%

Muon identification and measurement If enough shielding is provided expect reconstruction efficiency and momentum resolution not to deteriorate much

Forward jet-tagging and central veto Essential handle to increase S/N for WW and ZZ fusion processes Performance can be significantly degraded – though algorithms could be optimized

Trigger High thresholds for inclusive triggers; use of exclusive triggers selecting specific final states.


Calorimeters: CMS ECALCalorimeters: CMS ECALCrystals

Barrel: OK

Endcap : 3krad/hr at y=2.6

Further studies at high dose rates, long term irradiation

PhotosensorsBarrel: APDs – higher leakage current a higher noise

~100 MeV/ch

Endcaps: VPTs – R&D: on new devices may be needed

ElectronicsBarrel: OK

Endcap: R&D: More rad-hard electronics at |y|~3?

Activation: in endcaps reach several mSv/h – access will be difficult


Calorimeters: ATLAS LArCalorimeters: ATLAS LArSpace Charge Effects

GeV/cm2/sComfortable margin in

Barrel. Inner parts of em endcap and FCAL may be affected

HV Voltage DropComfortable margin in

Barrel. Small ‘wheel’ of em endcap sees a large current

Precision meas. not possible

Electronics: Probably OK? R&D: Use of another cryogenic liquid, with less charge

deposited per GeV, or a cold dense gas to address issues of space-charge and HV voltage drop

Critical density


Muon SystemMuon SystemCurrent ATLAS/CMS muon systems designed with safety factor of 3-5 w.r.t. background estimations (establish real safety margin once LHC operates)

Strong geometric dependence of radiation rates ,

Possible strategy:

extra shielding at high |y| reduces background everywhere

restrict high |y| limit of muon acceptance

Radio-activation at high |y| of shielding, supports and nearby detectors - may limit maintenance access

Balance super robust detectors vs shielding and reduced high- |y| acceptance

R&D: Study limit of current detectors - use of CSCs in barrel,

at high- |y| - higher rates – use straw chambers? MSGCs/GEMs?

3| |~ e


Level-1 TriggerLevel-1 TriggerTrigger Menus

Triggers for very high pT discovery physics: no rate problems – higher pT thresholds Triggers to complete LHC physic program: final states are known – use exclusive menus Control/calibration triggers with low thresholds (e.g. W, Z and top events): prescale

Impact of Reduced Bunch Crossing Period Advantageous to rebuild L1 trigger to work with data sampled at 80 MHz Could keep some L1 trigger electronics clocked at 25 ns Require modifications to L1 trigger and detector electronics

R&D Issues Data movement is probably the biggest issue for processing at 80 MHz sampling Processing at higher frequencies and with higher input/output data rates to the processing elements. Technological advances (e. g. FPGA ) will help Synchronization (TTC) becomes an issue for short x-ing period


DAQDAQContinuous and extraordinary evolution of computing and communication technologies – monitor the evolution of:Readout Network

Follow LHC machine luminosity – exploit parallel evolution of technologies main building block of DAQ is the switch – interconnecting data sources (event digitizers) and processing nodes (event filters) rapid progress in interconnection technologies started recently – LHC needs cannot yet be satisfied using a completely off-the-shelf system Technology Tracking

Complexity Handling Online computing systems will have ~ 10000 CPUs, issues of hardware and software management, reliability,remote access, security, databases Technology Tracking (e.g. those found in ISPs)

R&D: How to handle bandwidth (rate size) Bandwidth is an issue both for readout and for event building

Documents

LHC Symposium - May 3, 2003 1 Super LHC - SLHC LHC Detector Upgrade Dan Green Fermilab