Helicity of W bosons in Top Quark Decays at

CDFShulamit Moed, University of

GenevaOutline:

Introduction

Motivation

Overview of W helicity studies

1D measurement of W helicity fractions with 955pb-1 of data

2D measurement of W helicity fractions with 955pb-1 of data

Summary

2

Why is top quark interesting?

Youngest member of the SM family

Striking characteristic: HUGE mass , at EWSB scale (Yukawa coupling ~1) - what does this tell us?

Unique opportunity to probe bare quark properties (spin? charge?)

Top special relation to the Higgs boson

Is top the gateway to new physics?

Top chargeTop spinTop lifetimeTop mass

Branching ratiosRare decaysNon-SM decaysDecay kinematicsW helicityAnomalous coupling|Vtb|

Production cross sectionResonance productionProduction kinematicsSpin polarization

3

The Tevatron

Main Injector

Tevatron

DØCDF

Chicago

p source

Booster

W helicity result

with 955 pb-1

p-pbar collisions with 1.96TeV

center-of-mass energy.

Until LHC turns on - the only place to study top quark

4

Top production at the Tevatron

q

q

t

t

t

t

g

g

Top pair production

Main mechanism for top physics at Tevatron

Single top production Not yet observed different final states than pair production Larger background

For top mass = 175 GeV

@ √√s= 1.96 TeVs= 1.96 TeV: pb

pblumisyststat

tt

tt

8.07.6

)(4.0.)(6.0.)(5.03.7

~85% ~15%

(theory)

CDF combined

~1 top event every 10 BILLION inelastic collisions

L

NN bkgobs

tt

geometric and kinematic acceptance

selected eventsestimated bkg

5

t-tbar Final States

Top decays before hadronizing

~10-25 s (due to large mass)

Vtb~1; Mtop>MW+Mb: Decays to real W

BR(tWb) ~ 100%

all-jets44%

dilepton5%

lepton+jets15%

lepto

n+

jets

15

%e jets

e

jets

Final states are classified by the decay of the W’sBR(Wl) = 1/3BR(Wqq) = 2/3In all cases, the final state has 2 b quarks

Lepton+jets

Di-lepton

All hadronic

6

Lepton+Jets Channel

Best compromise higher statistics than dilepton, less background than all-hadronic Purity of the l+jet channels is not that high (S:B ~1:3)

Increase S:B by using b-tagging Fully reconstruct the event

Detected objects for full event reconstruction:

4 energetic jets

1 isolated charged lepton

Missing Et for the neutrino

Do not know with certainty the correct assignment between parents and decay products

??

7

Muon system

CMP ; CMU |η|<0.6

CMX 0.6<|η|<1.0

The CDF detector

Tracking system: Silicon detector -> b tagging COT : central outer trackerEff. for charged particle tracks: ~100% for |η|<1.0 ~40% for |η|≈ 2.0

calorimeters

Excellent lepton ID:~80% eff. for central electrons~90% eff. for high Pt muons

Up to |η|<3.6

z

x

y)2/tan(ln

pseudorapidity

8

W helicity in top quark decays

bWVt

igtb

5122

PJH

SM top decays via the weak interaction

V-A coupling like all other fermions:

t

spin=1/2

W

spin=1

b

spin=1/2

Helicity:

22

2

0

0

00

2

)()()(

)(

tw

t

RL

mm

mf

WWW

Wf

The longitudinal fraction:

This measurement:

Test of the SM, non-zero V+A?

EWSB – prediction of high longitudinal W fraction

9

What do we measureWhat do we measure? ?

bl

blbl

pp

EEpp )cos( *

We fully reconstruct the event using:

SM prediction of helicity fractions )assuming Mt=175GeV(:

longitudinal f0 = 0.7

left-handed f- = 0.3

right-handed f+ = 0

Left-handed longitudinal Right-handed

2* )cos1( 2* )cos1( )cos1( *2

V+A is suppressed

1/2

10

Sensitive Variables

cos(θ*)…

M2lb = 1/2 (M2

t – M2W)(1 + cos θ*)

Lepton-b invariant mass

Lepton transverse momentum

comp

lexity

11

Other W Helicity Measurements

LCF

F

.%95@27.0

74.0 22.034.00

RunI (Mlb)2 :

Early RunII:

1)7109( pbIntegrated luminosity=

Previously at CDF

D0 09/2006 370 pb-1

cos(θ*) method

LCf

f

AV

AV

.%95@23.0

057.008.0056.0

12

Analysis Overview

lepton+jets selection fully reconstruct the leptonic top

decay. calculate cos(θ*) construct templates for left-handed,

right handed and longitudinal W’s and background

fit helicity fractions using unbinned likelihood fitter.

correct for acceptance effects. estimate systematic uncertainties.

13

Event Reconstruction

Selection main features: only one isolated lepton with PT >20 GeV

at least 4 jets with ET > 15 GeV and |η|<2.0 (JETCLU with ΔR=0.4)

missing ET > 20 GeV

at least one jet is tagged with a secondary vertex tagging

veto on electrons from photon conversion

veto on events tagged by cosmic ray tagger

scalar sum of transverse energies of all reconstructed objects (Ht) > 200 GeV

Reconstructed objects – 4 jets + 1 lepton

24 permutations of possible combinations , which one we choose?

14

b-jet Tagging

Expect t W bb jet tagging is a very important tool.

- Every ttbar event contains 2 b-jets - Less than 20% of the dominant

background (W+jets) contains Heavy Flavor (b/c quarks)

B decay signature: displaced vertex Long life time c ~ 450 m: travels

Lxy~3mm before decaying

Require at least 1 jet tagged with the secondary vertex tagging algorithm.

Reduce permutations from 24 to 12!

b-tagb-tag

b-tagb-tag1.2 cm

CDF Event:CDF Event:

Close-up View of Layer 00 Silicon Close-up View of Layer 00 Silicon DetectorDetector

MET

15

Jet Energy Scale

Corrections applied to estimate the original parton energies from the observed jet energy in the calorimeter

Jets are corrected for:

η dependence correction – homogenous calorimeter response.

subtraction of energy due to pile-up of multiple interactions in the same bunch crossing.

correction for non-linearity and energy loss in the uninstrumented regions of the detectors.

Underlying event energy that falls inside the jet cone.

Jet energy radiating out of the jet cone.

Top specific corrections – flavor and topology of ttbar events.

16

Kinematical Fit

Provides constraints on W mass, t mass= anti-top mass etc.

Fit lowest Χ2 used to select the most likely combination

Χ2

2

2

2

2

2

2

2

2

4,

2,,2

t

tbl

t

tbjj

W

Wl

W

Wjj

jetsli i

measiT

fitiT MMMMMMMMPP

PT resolutions 1.5 GeV

2.5 GeV

MW , Mt = pole masses

efficiency ≈ 33%

17

Selected Data SampleSelected Data Sample

Use lepton+jets selection with at least one b-tagged jet and Ht>200GeV cut )to reduce QCD background( for events with njets >= 4:Data 220 events (89% signal fraction)

Total background 22.8 events

Process bkg events

fraction fraction

Mistag 9±1.35 39.5% 4.1%

W+h.f. 6.4±1.85 28% 2.9%

Single top 0.54±0.17 2.4% 0.25%

Diboson 1.36±0.07 6% 0.61%

QCD 5.5±1.08 24.1% 2.5%

Background composition

Scaled to 955pb-1

18

Monte Carlo Samples

Used HERWIG based samples with top mass of 175 GeV.

In these samples one the leptonic W is forced to a specific helicity (longitudinal, left handed or right handed).

The hadronically decaying W decays according to SM.

reconstructed

particle-level

19

ParameterizationsParameterizations

Comparison of signal and background fits used for likelihood fitter parameterizations.

Fit to 3rd order polynomial times exponential.

Background model is a mix of Wbbpp, W4p and diboson sample.

longitudinal

background

left-handed

right-handed

20

The LikelihoodThe Likelihood

sN

isbbbbsbb pfpfbsPbGL

1

** ))(cos)1()(cos()|(),|(

Fitter to extract helicity fractions:

Use unbinned maximum likelihood fit. Pseudo experiments generated from ideal functions look fine. Tested ‘real’ pseudo experiments with arbitrary f0, f-, f+. (from templates) Tested ‘real’ pseudo experiments SM ttbar sample. (pythia) Fit residual and pull width look fine.

Gaussian bkg constraint

Poisson probability for number of observed events

shape information

pFFpFpFps )1( 000

longitudinal

left-handed

right-handed

Extract two results by fitting for:

F0 while F+=0

F+ while F0 is fixed to the SM value @Mt=175GeV

21

The likelihood Fitter – Linearity ChecksThe likelihood Fitter – Linearity Checks

MC input(f0, f+)

Fits(f0, f+)

(0.3, 0.0)(0.310.03, fixed)

(0.5, 0.0)(0.510.03, fixed)

(0.7, 0.0)(0.710.03, fixed)

(0.9, 0.0)(0.900.03, fixed)

(0.7, 0.0)(fixed, 0.00 0.001)

(0.7, 0.1)(fixed, 0.100.01)

(0.7, 0.2)(fixed, 0.200.01)

Expected stat. uncertainty for f0 : δf0 = 0.12 and for f+: δf+ = 0.058

Linearity checks and sensitivity with <S+B>=220 events: All fits are consistent with “no bias”.

22

The Likelihood Fitter – SM exampleThe Likelihood Fitter – SM example

Right-handed,

f0 fixed

Means and pulls for realistic pseudo experiments constructed with SM values:

Longitudinal,

f+ fixed

f+ = 0

F+ = 0.006

f0 = 0.7

F0 = 0.708

σ = 10.023σ = 0.990.023

23

Acceptance

In the fitter, all templates are normalized to 1, In the fitter, all templates are normalized to 1, butbut– –

our acceptance for WL≠ W0≠ WR

Cuts on lepton PT and isolation left (right) acceptance is smaller (larger) relative to the longitudinal W’s.

Applying an acceptance correction:

Recall:

24

Acceptance Correction

fff

fF

00

000

)1(

)1(1

)1()1(

).(

).(

0

00

0

00

0

RfR

fF

RF

FRFcorrection

allongitudinAcc

handedleftAccR

F0=measured fraction ; f0= true fraction

for the right-handed fraction:Correction for f+ is very small (~0.01) not applied.Instead – assign a 1% systematic.

αi = accptance for helicity i

25

Systematic UncertaintiesSystematic Uncertainties

We use realistic pseudo experiments to estimate systematic uncertainties while keeping the fit unchanged.

Source

Bkg model

JES

Signal model

PDF

ISR/FSR

MC statistics

Instantaneous luminosity

Lepton energy scale

Acceptance correction

Total syst.

δf0

±0.038

±0.013

±0.020

±0.009

±0.010

±0.020

±0.007

±0.001

±0.001

±0.053

δf+

±0.017

±0.010

±0.010

±0.006

±0.005

±0.010

±0.002

±0.002

±0.001

±0.027Expected stat. uncertainty 0.12 0.06

26

Results – Data FitResults – Data Fit

Fitting the data:

F0 = 0.65 (measured)

f0 = 0.60 ± 0.12 ± 0.06, (corrected) f+ = 0 fixed

f+ = -0.06 ± 0.06 ± 0.03, f0 fixed to SM value

@Mt=175 GeV

Fit for right-handed fraction

Fit for longitudinal fraction

27

Results - Likelihood CurvesResults - Likelihood Curves

For longitudinal fraction

For right-handed fraction

28

Results – Setting Upper Limit on fResults – Setting Upper Limit on f++

Bayesian method for setting a limit@95% C.L:

Model systematic uncertainties as a gaussian with =0, σ= 0.027 .

- Have verified f+ systematic independent of f+

Convolute with likelihood

- as expected the effect is small, dominated by statistics.

f+<0.11@95% C.L

W systematics

w/o systematics

29

Expected Statistical Uncertainty

Assuming no improvements, stat~syst with 4fb-1.

30

2D Fit – First Simultaneous f0, f+ measurement !

000

0

0

0

11 FF

F

f

000

0

0

11 FF

F

f

Same data, same reconstruction, same templates etc.

fit for f0 and f+ simultaneously, rather than:

Fixing f+ to 0 (=SM) and fitting for f0

Fixing f0 to 0.7 (=SM) and fitting for f+

---> Less precision, but a more general result

when fixing one fit parameter to its SM value (1D fit), the correction is either simple (f0) or negligible (f+)

With the increasing luminosity:

Interest in a model independent measurement

V+A coupling bounded by CLEO bsγ data at a level that cannot be reached even at the LHC.

No assumption on helicity fractions while fitting, probe any deviation from SM (super-symmetry, dynamical electroweak symmetry breaking models, Extra dimensions ….)

31

Uncertainties for Simultaneous Fit

Systematics

statistical

0.25

0.10

Compared with 1D fit –

0.053 for f0

0.027 for f+

Compared with 1D fit –

0.12 for f0

0.06 for f+

Expected sensitivity from 1000 SM p.e:

32

2D Fit Results

f0 0.740.25(stat)0.06(syst)

f 0.060.10(stat)0.03(syst)

33

Limit on f+?

Form probability surface Find contour of constant

probability that captures 95% of the volume under the surface

No systematics in likelihood shape.

but for 2D fit: stat syst = stat

34

Summary

Results: 1D fit

f0 = 0.6 ± 0.12(stat.) ± 0.06(sys.), f+ = 0 fixed

f+ = -0.06 ± 0.06(stat.) ± 0.03(sys.), f0 fixed to SM

value @175GeV

f+ < 0.11 @ 95% C.L (including systematics)

2D fit

First simultaneous measurement of right-handed and longitudinal W helicity fractions! Improvement of CDF 1D results of longitudinal and right handed W fractions. Our knowledge of t-W-b vertex is still statistically limited. CDF now factor of 2 better than previous measurements. However still factor of 2 above current systematics - This is worth doing as a 4 fb-1 analysis on CDF. Measurement consistent with SM predictions – top decay is of V-A nature. Current status - working towards a publication .

f0 0.740.25(stat)0.06(syst)

f 0.060.10(stat)0.03(syst)

35

Back up slides

36

Top Mass Dependence

Top mass is not constrained in this analysis.

Fit to a linear function yields a correction of 0.5% for a 1σ variation of the top mass (3 GeV).

37

Systematic Uncertainties – BackgroundSystematic Uncertainties – Background

Background shape systematic:

Assume 100% W4p or 100% Wbb2p

Add 25% special QCD sample

Vary q2 for W sample

reminder - estimated ~5 QCD events out of 220

Special QCD sample Multi-jet trigger

0.8<em<0.95

Ntracks>3

38

Background Dominated SamplesBackground Dominated Samples

Comparison of 0-tag sample and bkg model

We have a reasonable background model

39

Systematic Uncertainties – MC StatisticsSystematic Uncertainties – MC Statistics

Statistical uncertainty of the parameterizations is not propagated through the analysis systematic uncertainty:

Re-fitting templates 1000 times, Poisson fluctuate the bins around central value.

Draw pseudo-experiments from the different fits.

Take difference in RMS of fitted values as a systematic :

δf0=0.02 δf+=0.01

40

Systematic Uncertainties - PDF

difference between MRST72 and CTEQ5L.

difference between MRST75 and MRST72.

variation of the 20 CTEQ6M eigenvectors.

21

220

1

))]()([(2

1

i

ii SFSFF

41

2D Pull Distributions

42

955pb-1 – Data/MC Comparison

43

955pb-1 – Data/MC Comparison