Associated b production Associated b production in with a dedicated in with a dedicated

triggertriggerMario Campanelli/ Michigan State Mario Campanelli/ Michigan State

UniversityUniversity

Experimental techniques:

Tevatron and CDF

Triggering on b: SVT

Heavy flavor measurements

•Low-pt

•High-pt

Thanks to:

A.Clark, X.Wu, S.Vallecorsa, T.Hoffmann, R.Lefevre, K.Hakateyama etc.

What’s interesting in HF production at What’s interesting in HF production at colliderscolliders

kHz rates at present Tevatron kHz rates at present Tevatron energy/luminosityenergy/luminosity

High mass -> well established High mass -> well established NLO calculations, NLO calculations, resummation of log(pT/m) terms (FONLL)

New fragmentation functions from LEP data

Gluon splittingFlavor creation

Leading Order Next to Leading Order

Flavor excitationg

g

g

g

Q

Q

other radiative corrections..

Release date of PDF

bN

LO(|y

|<1)

(b

)

<1994

now

B production mechanism B production mechanism and angular correlationsand angular correlations

The various b production The various b production mechanisms result in mechanisms result in different angular different angular distributions of the bb pairdistributions of the bb pair

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

Heavy-quark Pdf Heavy-quark Pdf uncertaintiesuncertainties QCD: sensitive to Pdf of b extrapolated QCD: sensitive to Pdf of b extrapolated

from gluon (high uncertainty at high x, from gluon (high uncertainty at high x, region probed at Tevatron)region probed at Tevatron)

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

Large uncertainties!

We need better than this for the LHC!

The experimental The experimental challengechallenge

b production 3-4 orders of b production 3-4 orders of magnitude smaller than magnitude smaller than ordinary QCD; selected by ordinary QCD; selected by longer lifetimelonger lifetime

impact parameter

Decay Length

Primary Vertex

Secondary Vertex

b/c

Traditional strategy: take unbiased prescaled triggers, identify b off-line

•At low-pt prescales are enormous: only decays with leptons are available

•New experimental techniques can largely help

Et (GeV)Et (GeV) L1 psL1 ps L2 psL2 ps

55 20/5020/50 300/1000300/1000

2020 20/5020/50 12/2512/25

5050 20/5020/50 11

7070 11 88

100100 11 11

The The TevatronTevatron World’s largest hadron colliderWorld’s largest hadron collider

First large-scale super-conducting First large-scale super-conducting magnet (4.2 T) acceleratormagnet (4.2 T) accelerator

6.28 Km length, theoretical 6.28 Km length, theoretical maximum about 1.4 TeV per beammaximum about 1.4 TeV per beam

√√s = 1.96 TeVs = 1.96 TeV

Started operation in 1987 (run 0), Started operation in 1987 (run 0), then collected about 100 pbthen collected about 100 pb-1 -1 until until 1996 (run I), then a long shutdown 1996 (run I), then a long shutdown until 2000, and Run II between 2001 until 2000, and Run II between 2001 and 2009and 2009

Accelerator upgrade Accelerator upgrade between Run I and Run IIbetween Run I and Run II

The Tevatron in itself was designed for much higher The Tevatron in itself was designed for much higher luminosities, limitations to the design were coming luminosities, limitations to the design were coming from the injection, where the old main ring (in the from the injection, where the old main ring (in the same tunnel) was used same tunnel) was used

Main ring removed

External main injector built

Run II luminosity Run II luminosity evolutionevolution

Highest initial Lum store: 2.90eHighest initial Lum store: 2.90e3232

Best integrated Lum week: 45 pbBest integrated Lum week: 45 pb-1-1

Best integrated Lum month: 165 pbBest integrated Lum month: 165 pb-1-1

Almost 3 fbAlmost 3 fb-1-1 already collected already collected Shutdown foreseen on October ’09Shutdown foreseen on October ’09

Luminosity projectionsLuminosity projections

9/30/03 9/30/04 9/30/05 9/30/06 9/30/07 9/30/08 9/30/09

30 mA/hr

25 mA/hr

20 mA/hr

15 mA/hr

Inte

gra

ted

Lu

min

osi

ty (

fb-1)

9

8

7

6

5

4

3

2

1

0

We are here now

By end FY07

By end FY09

likely

CDF II CDF II detectordetector

CalorimeterCalorimeter CEM lead + scint 13.4%/√ECEM lead + scint 13.4%/√Ett2%2% CHA steel + scint 75%/√ECHA steel + scint 75%/√Ett3%3%TrackingTracking (d0) = 40(d0) = 40m (incl. 30m (incl. 30m beam)m beam) (pt)/pt = 0.15 % pt (pt)/pt = 0.15 % pt

CDF fully upgraded for Run CDF fully upgraded for Run II:II:

Si & trackingSi & tracking New endcap calorimetry New endcap calorimetry L2 trigger on displaced tracksL2 trigger on displaced tracks High rate trigger/DAQHigh rate trigger/DAQ

Online trackingOnline tracking The SVT is a The SVT is a

revolutionary device revolutionary device made of 150 VME made of 150 VME boards that measures boards that measures track parameters in a track parameters in a 15 15 s pipeline. Its s pipeline. Its results can be used at results can be used at Level 2 triggerLevel 2 trigger

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

1.1. Find low resolution track (road) in COTFind low resolution track (road) in COT2.2. Discretize Discretize , Pt of road and SVX hits, Pt of road and SVX hits3.3. Compare with pre-calculated configurations Compare with pre-calculated configurations

(in associative memory): no fit performed!(in associative memory): no fit performed!4.4. Track parameters found in few Track parameters found in few ss

Performances of online Performances of online trackingtracking

Hit finding efficiency: 85%Hit finding efficiency: 85% Impact parameter Impact parameter

resolution: resolution:

This device has greatly changed the way we do b physics at CDF, both at low and high Pt.

In the rest of this talk I will present examples of measurements of b properties, and how the presence of on-line tracking improved them or made them possible

35 m 33 m = 47 m (resolution beam)

Low-Pt B Physics Low-Pt B Physics TriggersTriggers

Di-muonJ/->B ->

Two muons with:pT() > 1.5 GeV/c

One displaced track +lepton (e, )

B -> lXLepton:

pT(l) > 4.0 GeV/cTrack:

pT > 2.0 GeV/c,

d0 > 120 m

Two displaced tracksB -> hh

Two tracks with:pT > 2.0 GeV/cpT > 5.5 GeV/c

d0 > 100 m

Completely newLargely improved

Low-pt b physics in CDFLow-pt b physics in CDF Impact parameter trigger responsible for an Impact parameter trigger responsible for an

enormous wealth of results on b physics:enormous wealth of results on b physics: Bs oscillationsBs oscillations B->hh modesB->hh modes bb discovery discovery SpectroscopySpectroscopy LifetimesLifetimes ..and many others..and many others

Exclusive/low Pt Exclusive/low Pt measurements on measurements on charmed mesonscharmed mesons

Cross section of exclusive Cross section of exclusive charm statescharm states

With early CDF data:With early CDF data: 5.8 5.80.3pb0.3pb-1-1

• Measure prompt charm meson

production cross section

• Data collected by SVT trigger

from 2/2002-3/2002

• Measurement not statistics limited

Large and clean signals:

Need to separate direct D and BD decay

• Prompt D point back to collision point

I.P.= 0

• Secondary D does not point back to PV

I.P. 0

Separating prompt from Separating prompt from

secondary Charmsecondary Charm

Direct Charm Meson Fractions:Direct Charm Meson Fractions:

DD00: f: fDD=86.4±0.4±3.5=86.4±0.4±3.5%%

D*D*++: : ffDD=88.1±1.1±3.9%=88.1±1.1±3.9%

DD++: f: fDD=89.1±0.4±2.8%=89.1±0.4±2.8%

DD++ss: f: fDD=77.3±3.8±2.1%=77.3±3.8±2.1%

BD tail

Detector I.P. resolution shape measured

from data in K0s sample.

Most of reconstructed charm mesons are direct

Separate prompt and secondary

charm based on their transverse

impact parameter distribution.

Prompt D Secondary D from B

Promptpeak

Single Charm Meson X-Single Charm Meson X-Section Section

Calculation from M. Cacciari and P. Nason:

Resummed perturbative QCD (FONLL)

JHEP 0309,006 (2003)

CTEQ6M PDF

Mc=1.5GeV,

Fragmentation: ALEPH measurement

Renorm. and fact. Scale: mT=(mc2+pT

2)1/2

Theory uncertainty: scale factor 0.5-2.0

PT dependent x-sections:

Theory prediction:

With more data: charm pair With more data: charm pair correlationscorrelations

Single D meson cross Single D meson cross section already section already systematics dominated; systematics dominated; more statistics useful to more statistics useful to look for angular look for angular distributions and distributions and discriminate production discriminate production mechanismsmechanisms

As expected, data favor more gluon splitting than LO Monte Carlo. Use of higher orders should improve the agreement

High-Pt measurements: b High-Pt measurements: b production on unbiased production on unbiased

datasetsdatasets

Absolute jet correctionsAbsolute jet corrections

Systematics for energy Systematics for energy scalescale

Major systematics for most of jet-based measurements

B-jet identification: search B-jet identification: search for secondary vertexfor secondary vertex

For inclusive studies, instead For inclusive studies, instead of trying to identify specific b of trying to identify specific b decay products, we look for a decay products, we look for a secondary vertex resulting secondary vertex resulting from the decay of the b mesonfrom the decay of the b meson

Efficiency of this “b Efficiency of this “b tagging” algorithm tagging” algorithm (around 40%) is taken (around 40%) is taken from Monte Carlo and from Monte Carlo and cross-checked with b-cross-checked with b-enriched samples (like enriched samples (like isolated leptons)isolated leptons)

b-jet fractionb-jet fractionWhich is the real b content Which is the real b content (purity)(purity)??Extract a fraction directly from dataExtract a fraction directly from data Use shape Use shape secondary vertex masssecondary vertex mass

Different PDifferent Ptt bins to cover wide spectrum bins to cover wide spectrum Fit data to MC templatesFit data to MC templates

MonteCarlo templatesb

non-b

98 < pTjet < 106 GeV/c

High pHigh ptt b jet cross b jet cross sectionsection

MidPointMidPoint R Rconecone 0.7, |Y| < 0.7 0.7, |Y| < 0.7 Pt ranges defined to have Pt ranges defined to have 99% 99% efficiencyefficiency ((97% 97% Jet05)Jet05)

Jets corrected for det effectsJets corrected for det effects

~ 300 pb-1

Pt ~ 38 ÷ 400 GeV

20 ÷ 10%

Inclusive calorimetric triggersInclusive calorimetric triggers L3 Et > x L3 Et > x (5,20,40,70,100)(5,20,40,70,100)

Jets unfolded back to parton level for comparison with NLO cross section (Mangano et al. 1997)

Large uncertainties due to factorization and renormalization scales (default 0/2); overall data higher than theory in the high-Pt region (where gluon splitting is more present)

Possible need for higher orders

High pHigh ptt b-jet cross b-jet cross sectionsection

Using the SVT at high PtUsing the SVT at high Pt Since the beginning of Run II in addition to the low-Pt Since the beginning of Run II in addition to the low-Pt

triggers there are trigger paths that use SVT triggers there are trigger paths that use SVT information to enhance b content in high-Pt events.information to enhance b content in high-Pt events.

Conceived to search for new physics, we are now Conceived to search for new physics, we are now analyzing these datasets to measure QCD properties:analyzing these datasets to measure QCD properties: PHOTON_BJETPHOTON_BJET

A photon with Et>12 GeVA photon with Et>12 GeV A track with |dA track with |d00|>120 |>120 mm A jet with Et>20 GeV (SVT selects about 50% of tagged jets)A jet with Et>20 GeV (SVT selects about 50% of tagged jets)

HIGH_PT_BJETHIGH_PT_BJET 2 tracks with |d0|>120 2 tracks with |d0|>120 mm 2 jets with Et>20 GeV2 jets with Et>20 GeV

Z_BBZ_BB 2 tracks with |d0|>160 2 tracks with |d0|>160 mm 2 jets with Et>10 GeV (to have larger BG sidebands)2 jets with Et>10 GeV (to have larger BG sidebands)

bb cross sectionbb cross sectionBased on 260 pbBased on 260 pb-1-1 of data taken with of data taken with

SVT-based trigger- no prescaleSVT-based trigger- no prescale

In the absence of a control sample, efficiency computed from MC.

To avoid detailed trigger simulation, offline cuts harder than trigger requirements.

Efficiency about 2% vs 10% of just requiring 2 b-tags

SVT requirement « costs » about a factor 5 (eff. ~50% per jet)

B purity around 90% due to double SVT + offline tag requirement

bb cross sectionbb cross sectionLargely dominated by systematics

Dominant systematic uncertainties:Jet Corrections 15-20%b fraction 7%Unfolding 4%

Angular distributionAngular distributionConfirm excess of events

with small angle between the jets wtr leading order MC

Proper treatment of multi-parton interactions is as important as inclusion of next-to-leading order

B jet energy scale using Z-B jet energy scale using Z->bb>bb Apart from being an interesting challenge per se, finding the Z Apart from being an interesting challenge per se, finding the Z

in the hadronic b decay channel is useful for jet energy scale in the hadronic b decay channel is useful for jet energy scale calibration. Start from very large SVT-based di-jet datasetcalibration. Start from very large SVT-based di-jet dataset

Dijet Dijet event event selectionselection

Low jet ELow jet Ett

Z signal Z signal distinct distinct from BG from BG peakpeak

Two leading Two leading R=0.7 jets:R=0.7 jets:

• EEtt(L5)>22 (L5)>22 GeVGeV

• ||detdet|<1.0|<1.0

Signal Signal optimizatiooptimizationn

Back-to-Back-to-back back events events with low with low extra jet extra jet radiationradiation

(jet 1, jet (jet 1, jet

2)2)>3.0>3.0

EEtt(3rd jet)(3rd jet)

(corr)<15 (corr)<15 GeVGeV

Tagging Tagging both jets are both jets are (+) SecVtx (+) SecVtx taggedtagged

BG BG modelingmodeling

Data-drivenData-driven

ET3

12

[ΔΦ12]BG

[Et3]BG15

3.0

Tag ratio: BG Region

Signal Region

BG Template

Dijets mass

Taggable events

Fraction of (++) events vs dijet mass

Data-driven BG Modeling

Distribution is then parameterized with a p.d.f.

1

2

1 2

Z->bb signalZ->bb signal

b-JES = 0.974 ± 0.011(stat.) + 0.017 - 0.014 (syst.)

Nevents = 5674 ± 448(stat.) + 1473 - 570 (syst.)

Nexpected (Pythia) = 4630 ± 727

Background modeled as sum of Pearson and error function:

Signal: templates as a function of reconstructed Z mass (energy scale)

Photon + b Photon + b Analysis Analysis

Off-line cuts:Off-line cuts: |||<1.0|<1.0 jet with secondary vertexjet with secondary vertex Determine b, c, uds Determine b, c, uds

contributions contributions Subtract photon Subtract photon

background using shower background using shower shape fitsshape fits

Use Et > 25 GeV Et > 25 GeV unbiased photon dataset, without jet requirements at trigger level:

Photon Photon identificationidentification

No event-by-event photon No event-by-event photon identification possible: only identification possible: only statistical separation based statistical separation based on shower shape measured on shower shape measured by tracking devices inside by tracking devices inside the calorimeterthe calorimeter Pre-shower

Detector (CPR)

Central Electromagnetic Calorimeter

Shower Maximum Detector (CES)

Wire Chambers

SVT vs unbiased datasetSVT vs unbiased dataset

Two datasets can be used:

•Unbiased with Et()>25

•SVT-based with Et()>12 GeV without prescale!

Hard to calculate

Requires trigger simulation

Trigger efficiency can be computed directly from data using the overlap region (ISOSVT), where events have photon Et above 25 GeV and an SVT track

Analysis strategyAnalysis strategy

What we need in the end is:

• perform the analysis on the unbiased dataset

• do the same on SVT-based events with E(L3)>25 GeV

• use their ratio to extrapolate to the low photon Et region.

This only requires the hypothesis that the jet quantities are independent on the photon energy

Trigger efficiency: Et Trigger efficiency: Et dependencedependence

Trigger efficiency was found to be stable with run number, for the different trigger conditions.

Trigger efficiency is also constant as a function of jet Et (small dependence used in the analysis)

Photon + b Photon + b result on result on unbiased datasetunbiased dataset

Cross sections and ratio agree with LO predictions from MC, but on the high side

Photon+b cross Photon+b cross section on the SVT section on the SVT

datasetdataset

Data:

90.5 ± 6.0 (stat.) +21.7 -15.4 (syst) pb

Pythia gen. Level: 69.3 pb

Very good agreement with previous measurement based on PHOTON_25_ISO (~45% candidate overlap)

•Luminosity: 6%

•Trigger efficiency extrapolation (from statistics):10%

•Jet energy scale: 4% (from JES group methods)

•B purity templates: +20% -10%

bb + photon bb + photon productionproduction

Natural follow-up of the b-photon analysis, using the same dataset and analysis technique, requiring double b-tag

•trigger efficiency is 60%

•bb purity around 80%

Measurement still preliminary;

Performed on the full 1.2 fb-1 sample

About 1 event/pb-1

Angular distributions sensitive to production mechanisms

ConclusionsConclusions Sometimes a breakthrough in hardware Sometimes a breakthrough in hardware

technologies opens up a new world of physics technologies opens up a new world of physics opportunitiesopportunities

The possibility of online tracking allowed CDF to The possibility of online tracking allowed CDF to explore an exciting new horizons in heavy-flavor explore an exciting new horizons in heavy-flavor physicsphysics

The first analyses using SVT-based triggers at The first analyses using SVT-based triggers at high-Pt are ready, and show that we can high-Pt are ready, and show that we can understand these triggers well enough to perform understand these triggers well enough to perform precision measurementsprecision measurements

Now that the method has been tested on known Now that the method has been tested on known physics, SVT-based triggers start to be used for physics, SVT-based triggers start to be used for the purpose they were designed for- a new the purpose they were designed for- a new weapon to look for SUSY and Higgs (Hweapon to look for SUSY and Higgs (H++->bb ->bb analysis out soon)!analysis out soon)!