3000 fb-1 300 fb-1

Brian Petersen 12 November 2016 Higgs Couplings Higgs Program and Higgs Program and Detector Implications Detector Implications for HL-LHC for HL-LHC

Transcript of 3000 fb-1 300 fb-1

Page 1: 3000 fb-1 300 fb-1

Brian Petersen

12 November 2016Higgs Couplings

Higgs Program andHiggs Program andDetector ImplicationsDetector Implications

for HL-LHCfor HL-LHC

Page 2: 3000 fb-1 300 fb-1


IntroductionHiggs discovery highlightof LHC programDetailed measurementsof its properties in progress

So far appears consistentwith SM predictions

Precision measurementsare required to understandelectroweak symmetrybreakingHiggs could also be a portal to BSM physics

Strong motivationfor High-Luminosity LHC

arXiv: 1606.02266

Page 3: 3000 fb-1 300 fb-1


HL-LHC Upgrade

Page 4: 3000 fb-1 300 fb-1


HL-LHC Planning LHC to deliver 300 fb-1 by 2023 (end of Run-3)HL-LHC goal is deliver 3000 fb-1 in 10 years

Implies integrated luminosity of 250-300 fb-1 per yearRequires peak luminosities of 5-7x1034 cm-2s-1

while using luminosity leveling (3-5 hours at peak luminosity)Design for “ultimate” performance 7.5x1034 cm-2s-1 and 4000 fb-1

Page 5: 3000 fb-1 300 fb-1


HL-LHC Project

New IR-quads Nb3Sn(inner triplets)New 11 T Nb3Sn(short) dipolesCollimation upgradeCryogenics upgradeCrab CavitiesCold poweringMachine protection...

Major intervention on more than 1.2 km of the LHC

Machine upgrade approved by CERN council in June 2016

Page 6: 3000 fb-1 300 fb-1


The High-Luminosity Challenge

Very high pile-up Intense radiation

Major experiment upgrades needed to:Improve radiation hardness and replace detectors at end-of-lifeProvide handles for mitigating pile-up (high granularity, fast timing)Allow higher event rates to maintain/improve trigger acceptance

Goal is to maintain or improve over current performanceHiggs physics provides main benchmarks for this

HL-LHC provides an extreme challenge to the experiments

Page 7: 3000 fb-1 300 fb-1


Detector Upgrade – ATLAS Calorimeters

New BE/FE electronicsNew HV power suppliesLower LAr temperature

TrackerAll silicon tracker (strip and pixel)Radiation tolerant, high granularityLow material budgetCoverage up to |η|=4

Muon SystemNew BE/FE electronicsNew RPC layer in inner barrelMuon-tagging in 2.7<|η|<4.0(under study)

Trigger and DAQL0 rate at ~ 1 MHz(latency up to 10 μs)Possible hardwareL1 track triggerHLT output ~10 kHz

(Timing detector)High granularitytiming detectorCoverage: 2.5<|η|<4.2Possibly absorberfor |η|<3.2

Page 8: 3000 fb-1 300 fb-1


Detector Upgrades – CMS Endcap Calorimeter

High-granularity calorimeterbased on Si sensorsRadiation-tolerant scintillator3D capability and timing

Barrel CalorimeterNew BE/FE electronicsECAL: lower temperatureHCAL: partially new scintillatorPossibly precision timing layer

TrackerRadiation tolerant, high granularityLow material budgetCoverage up to |η|=4Trigger capability at L1

Muon SystemNew Be/FE electronicsGEM/RPC coverage in 1.5<|η|<2.4Muon-tagging in 2.4<|η|<3.0

Trigger and DAQTrack-trigger at L1(latency up to 12.5 μs)L1 rate at ~ 750 kHzHLT output ~7.5 kHz

Page 9: 3000 fb-1 300 fb-1


Higgs Physics at HL-LHC

Page 10: 3000 fb-1 300 fb-1


Higgs program at HL-LHCHiggs boson studies are a major component of HL-LHC physics programMain Higgs measurements at HL-LHC:

Higgs couplingsHiggs self-couplingRare Higgs decaysHiggs differential distributionsHeavy Higgs searches

HH ~100k

Higgs Production Channels

Higgs Decay Channels

3000 fb-1

3000 fb-1

Page 11: 3000 fb-1 300 fb-1


Physics ProjectionsHL-LHC Higgs prospects done in two ways:

Parameterized detector performanceEvent-generator level particles smeared with detector performance parameterized from full simulation and reconstruction of upgraded HL-LHC detectorsEffects of pile-up included for either 5x1034 cm-2s-1 (140 pile-up events) or 7x1034 cm-2s-1 (200 pile-up events)Analysis mostly based on existing 8 TeV analyses with simple re-optimization for higher luminosity

Extrapolation of Run-1 or Run-2 resultsScale signal and background to higher luminositiesCorrect for different center-of-mass energyAssume unchanged analysis (not re-optimized for higher luminosity)Assume same detector performance as in Run-1/2(some use corrections based on studies in first approach)

Page 12: 3000 fb-1 300 fb-1


Systematics TreatmentWith large statistics at HL-LHC, systematics can be dominating in measurement precision

Hard to predict how these will evolve with luminosity/timeBoth experiments start from current systematics with a slightly different approachATLAS approach:

Experimental systematics scaled to best guess for HL-LHCResults provided with current theory systematics and without theory systematics

CMS approach:Provide results in two scenarios:

Scenario 1: Current experimental and theory systematics

Scenario 2: Experimental scaled with luminosity (1/√L) until a certain best achievable uncertainty level The current theory systematics is halved

Both approach aim to bracket the achievable precision

Page 13: 3000 fb-1 300 fb-1


Projections for Higgs CouplingsFull set of HL-LHC coupling projectionsare based on Run-1 analyses

For μ=140 in case of ATLASSame as Run-1 performance for CMS

Higgs coupling precision (per experiment):3-5% for W, Z and γ~7% for µ5-10% for t, b and τ

Do not include improved detector designsor improvements in analysis techniques

300 fb-1 3000 fb-1



Page 14: 3000 fb-1 300 fb-1


Projections based on Run-2 AnalysisH→γγ and H→ZZprojections updated to 13 TeV (12.9 fb-1)based Run-2 analysesH→γγ added expecteddegradation at μ=200

Beamspot ~5cmVertex identificationreduced from 80% to 40%Photon ID efficiencydecreased by 2.3% (10%) in EB (EE)

Theory uncertaintiesbecome dominate at HL-LHCDecouple by measuringfiducial cross section

Can achieve ~4% precision

3000 fb-1300 fb-1


Page 15: 3000 fb-1 300 fb-1


Projections based on Run-2 AnalysisH→γγ and H→ZZprojections updated to 13 TeV (12.9 fb-1)based Run-2 analysesH→ZZ added expecteddegradation at μ=200

Reduced lepton efficiencyIncreased misidentification

Can make precisedifferential pT(H) cross sectionmeasurements

3000 fb-1300 fb-1

3000 fb-1300 fb-1



Page 16: 3000 fb-1 300 fb-1


Higgs Self CouplingMeasurement of Higgs pair productionmajor goal of HL-LHC program

Requires full HL-LHC luminosity to reach SM sensitivityAllows for a measurement of self coupling λ

Extremely challenging due to low cross section (~40 fb)


Page 17: 3000 fb-1 300 fb-1


Higgs Self Coupling Projections

Channel Expected limit in μ Significance Limits on λ/λSM

at 95% CL

Full Syst.

Stat. only

Full Syst.

Stat. only

Full Syst. Stat. only

gg→HH→γγbb 1.3σ -1.3<λ/λSM


gg→HH→ττbb 4.3 0.6σ -4<λ/λSM


gg→HH→bbbb 5.2 1.5 -3.5<λ/λSM

<11 0.2<λ/λSM




bbbb 0.35σ

CMS extrapolations from Run-2 analyses:

ATLAS simulations (HH→bbbb is Run-2 extrapolations):






Page 18: 3000 fb-1 300 fb-1


HH→bbbb AnalysisHH→bbbb analysis dominated by largemulti-jet background

Very difficult to simulateMultijet background is estimated from control regions (CRs)

Systematics uncertaintyassigned from CR differences

CR differences in Run-2 are smaller than the statistical uncertaintyBackground uncertainty should improve with more luminosity

By how much?

Run-2 m4j extrapolated to 3000 fb-1, 14 TeV



Page 19: 3000 fb-1 300 fb-1


Impact of H(125) on Detector Design

Page 20: 3000 fb-1 300 fb-1


Higgs Impact on Upgrade DesignThe design of the upgraded HL-LHC detectors is complex process

Want ultimate performance, but limited by what can reasonably be upgraded during long shutdown and by cost

Higgs measurements are corner stone of the HL-LHC physics program and they provide prime motivation for many upgrades beyond current detector capabilities

Will highlight four different cases:Pile-up jet suppression in VBF Higgs productionH→γγ reconstruction with precision timing detectorH→ττ triggeringHH→bbbb triggering and reconstruction

Page 21: 3000 fb-1 300 fb-1


Extended TrackersATLAS and CMS planto extend tracker coverageto η~4 with pixel extensionProvides multiple benefits

Extended lepton coverage(with forward muon tagger)Forward b-taggingImproved vertexing

Primary benefit is pile-up suppression


Page 22: 3000 fb-1 300 fb-1


Pile-up Jet SuppressionAt 200 pile-up, every events has ~5 pile-up jets (pT>30 GeV)Can suppress these by using tracking to associate them to either pile-up or hard-scatter vtxFor VBF Higgs productionneed to use jets out to η~4

Extended tracker enables thisJETM-2016-012 CERN-LHCC-2015-010


Page 23: 3000 fb-1 300 fb-1


VBF H→WW→eνμν Analysis

Different levelsof backgrounduncertaintieswith respect toRun-1 H→WWanalysis

Tracker coverage


Physics gain of forward trackerstudied in H→WW analysisSimple cut based analysis:

2 forward jets (|η|>2) inopposite hemispheresNo jet above 30 GeV in between jetse/μ in between forward jetsMissing ET>20 GeV

After selection: ~200 signal events~400 background events from tt and non Higgs WW

Factor two gain in precision from extended tracker coverage

Signal precision and significance


Page 24: 3000 fb-1 300 fb-1


High Granularity Timing DetectorMinimum biasscintillatorsHigh-granularity

timing detectorAdditional pile-up rejection canbe achieved using precise timing

Different time of flight anddifferent collisions times in event

ATLAS considering thin timing deviceFour layers silicon sensorsCoverage for 2.4<|η|<4.2Possible Tungsten absorber for |η|<3.2Timing target: 30-50 ps per MIP

Provide additional sensitivity to VBFPossibly also enhance the jet trigger


Page 25: 3000 fb-1 300 fb-1


Timing Detectors in CMSEndcap calorimeter (1.5<|η|<3) replaced by multi-layer silicon-based calorimeter

Current calorimeter not rad-hard enoughUse of silicon allows intrinsic time resolution down to 50 ps for large signalBarrel calorimeter electronics upgraded to also provide precision timing (30 ps)Additional timing layer for charged particles in front of calorimeter under consideration

Allows to reconstruct vertex time Example: Improved H→γγ vertex associationCMS-DP-2016-008


Page 26: 3000 fb-1 300 fb-1


H→γγ with Timing Detector

With full use of calorimeterand charged particle timinginformation vertexing efficiencycan be almost full recovered

Corresponds to effectively30% more luminosity

Vertex selection efficiency dropswith increase in pileup

~80% now → ~40% at 200 pileupResults in large degradation of mass resolutionImpact on fiducial crosssection measurement investigated

Page 27: 3000 fb-1 300 fb-1


Triggering on H→ττH→ττ channel critical for understanding fermionic coupling and measuring Higgs CP propertiesDifficult to trigger on efficiently

Two narrow, fairly soft jets with 1-3 charged tracksExisting calorimeter-only L1 triggers not sufficient

Acceptance drops quickly as thresholds are raisedAdding fast track trigger can give large rate reductionCMS estimate: 50 kHz L1 rate for 45% eff. for VBF H→ττ

Same triggers also useful for HH→bbττ CERN-LHCC-2015-010

Page 28: 3000 fb-1 300 fb-1


Triggering on HH→bbbbHH→bbbb channel also difficult to trigger on at L1

Very large rate of multi-jets and pile-up jetsPlan to also use track triggerto suppress pile-up jetsin 4-jet triggerStill likely to only beefficient at 70-75 GeVATLAS estimate this willreduce sensitivity by ~30%compared to current 30 GeV

Better trigger strategyis under investigation


Page 29: 3000 fb-1 300 fb-1


B-tagging for HH→bbbb


Efficient and highly rejecting b-tagging also critical for HH→bbbb measurement

Current projections assume performance as in Run-2Both experiments have demonstrated ability to matchcurrent performance at pile-up of 140 eventsBoth pixel detectors still being optimized

Aim to achieve Run-2 performance at pile-up of 200



Page 30: 3000 fb-1 300 fb-1



Page 31: 3000 fb-1 300 fb-1



Much more information in presentationsat HL-LHC Experiments workshop inAix-Les-Bains in October:https://indico.cern.ch/event/524795/timetable/

High-Luminosity LHC very challenging environmentMajor detector upgrades being planned to cope with itPrecision Higgs measurements are themain physics driver for detector upgradesExpect that upgraded detector in most areas can match current performance at highest pile-up levelsand will be even better in some areasTechnical Design Reports in preparationand will come over the next ~2 years

Please stay tuned

Page 32: 3000 fb-1 300 fb-1



Page 33: 3000 fb-1 300 fb-1


Anomalous HZZ Coupling

Test for anomalous HZZcouplings ai:

Interference contributionbecomes more dominant atsmaller values of fai x cos(φai)


Page 34: 3000 fb-1 300 fb-1


Higgs to Invisible

Main backgrounds:Z(ℓℓ)+jetsW(ℓν)+jetsQCD multijet

Current BR(H→inv) limit (expected):BR<0.30 @ 95% CL (CMS)BR<0.31 @ 95% CL (ATLAS)

Projected upper limit (CMS) asas function of luminosity:


Page 35: 3000 fb-1 300 fb-1


Search for ttHH

Selection with ≥5 b-tags:25 signal events, 7100 background eventsBackground dominated by c-jets from W mis-tagged as b

Significance for ttHH production without systematics: 0.35σ

σ(ttHH)~1fbUse HH→bbbb final state and semi-leptonic tt decaySingle lepton trigger requirementSignature: 6 b-jets, 2 light jets, lepton and missing energy

Cut-based analysisNo cuts on Higgs candidate mass due to combinatorics



Page 36: 3000 fb-1 300 fb-1


VBF H→ZZ*→ℓℓℓℓInitial selection:

2 jets with m(jj)>130 GeV4 leptons consistent with H→ZZ*→ℓℓℓℓ

Use BDR to separate ggF and VBFLarge pile-up contribution in ggF

190 signal events and330 background eventsResults with full systematics (signal QCD scale)and statistics only:


Page 37: 3000 fb-1 300 fb-1


Summary of Recent ATLAS Higgs Results

Page 38: 3000 fb-1 300 fb-1


Wanted Reduction in Theory UncertaintiesATL-PHYS-


Page 39: 3000 fb-1 300 fb-1


Search for Heavy Higgs→ττ

300 fb-1

3000 fb-1

One of the most sensitivechannels for constrainingextended HiggsCross section limits:

ggφ (→ττ)bbφ (→ττ)

Model dependent limits:mmod+ benchmark

Sensitivity at high mA isstill dominated by statistics

Page 40: 3000 fb-1 300 fb-1


CMS Tracker Changes

Page 41: 3000 fb-1 300 fb-1


CMS Tracker Comparison

Page 42: 3000 fb-1 300 fb-1


ATLAS Tracker Hits and MaterialOptimized for at least 13 hits,minimum material and coverageup to η=4



Page 43: 3000 fb-1 300 fb-1


CMS Tracker PerformanceCMS-DP-2016-065

Page 44: 3000 fb-1 300 fb-1


ATLAS Tracker PerformanceATL-PHYS-


Page 45: 3000 fb-1 300 fb-1


New CMS Endcap Calorimeter

Page 46: 3000 fb-1 300 fb-1


ATLAS Trigger Schemes

Level-0 + Level-1 hardware trigger Level-0 only hardware trigger

Page 47: 3000 fb-1 300 fb-1


CMS Trigger SystemCurrent Level-1 trigger uses only calorimeter and muon informationPhase-II upgrades

Replace calorimeter electronicsIncrease latency and Level-1 accept rateUse tracking at Level-1 based on doublet seedsGlobal track-trigger correlator

Page 48: 3000 fb-1 300 fb-1


ATLAS Example Trigger MenuFor most trigger channels, expect to maintain same or even lower trigger threshold as in Run-1

Hadronic triggers challenging due to pile-up

Page 49: 3000 fb-1 300 fb-1


CMS Example Trigger MenuMenu without track-triggerhas 1.5 MHz rate μ=140

Track-trigger givesfactor 5.5 reduction:260 kHzUse 1.5 safety factor:390 kHz

Menu with track-triggerhas 500 kHz rate μ=200

With 1.5 safety factor:750 kHzWithout track-trigger:~4 MHz

Page 50: 3000 fb-1 300 fb-1


CMS Precision Timing for Charged ParticlesAssume sufficient timing performance for charged hadrons,e.g. from dedicated LYSO+SiPM layer in the central region,and from HGCAL or dedicated layer in the forward regionTraditional three-dimensional vertex fit can be upgraded to afour-dimensional fit, with vertices reconstructed both in position along the beamline and in time within the bunch crossingProvides further suppression of charged particles from pile-up for jets, missing energy, lepton isolation etc

20 ps resolutionassumed for chargedparticles with p

T>1 GeV


Page 51: 3000 fb-1 300 fb-1


Pile-up vs Pile-up Density

Lepton isolation efficiency

B-tagging efficiency

So far mostly considered effects due to overall pile-upFind that many quantities depend more onpile-up density – how many in pile-up collisions per mm in zThis can be mitigated by changingbeam-profile

I.e. spreading vertices out better in z CMS-DP-2016-065