Transcript Top Quark Properties from CDF

Top Quark Properties
from CDF
Robin D. Erbacher
University of California, Davis
Fermilab Wine and Cheese -- Friday June 10, 2005
Top Quark Discovery: 1995
The search for top lasted almost two
decades. Its unexpectedly heavy mass
delayed discovery.
CDF + D0 combined:
Mass (top) = 178  4.3 GeV/c2
5 orders of magnitude
2
Why Is Top So Interesting?
Well, top physics is different!
 top ~ 10-24 s , 1  1.5 GeV
1
<< -1QCD ~ (200 MeV)-1
•Top quark lifetime is short: decays before hadronizing
No spectroscopy like other heavy flavorIn Top Color, the Higgs is
a bound state of top
Top momentum and spin transferred quarks (C. Hill)
to decay products
• Probes physics at higher scales
than other known fermions
Top (or heavy top) very hip in many
EWSB models: Higgs, Top Color,
Little Higgs, SUSY mirror models
3
Elucidating the Top Quark in Run 2
Vtb
Top pairs: (tt) ~7 pb
•Top production rate
•Mass of top
•W helicity in top events
•QCD tests
•New
in X tt
Newphysics
physics!
•Anomalous couplings,
new particles
Single top: (tb) ~3 pb
•|Vtb|
•QCD tests
•New
physics?
New
physics!
4
Number of Physicists
# of Physicists for Particle Discovery
CDF (Tevatron) ~ 800 (1500)
LHC
Year Discovered
5
Physics of the Top Quark
Top physics is still one of the more sexy things to
study at the Tevatron…
6
How is Top Produced?
~85%
q-q
Rarely!!
Standard Model
Tevatron Pair Production
Through Strong Interaction
~15%
g-g
 ( pp  tt @ M top  175GeV )  6.7 pb
One top pair each 1010 inelastic collisions at s = 1.96 TeV
7
How Else is Top Produced?
Standard Model Tevatron Single Top Production
 ( pp  t  X @ M top  175GeV )  3 pb
p
X
p
t
t
Resonance Production?
Top Color-Assisted Technicolor
OR
?????
8
How Does Top Decay?
Standard Model:
tWb ~ 100%
Main “usable” top event topologies:
• tt  llbb
di-lepton
5% e+
• tt  lqqbb lepton+jets 30% e+
• tt  qqqqbb all hadronic 45%
9
Identifying Top Quarks
What do we look for in top events?
 Electrons
 Muons
 Neutrinos
 Quarks (Jets)
 b Quark Jets
“All
Hadronic
Channel”
“Lepton
+
Jets
Channel”Analyses!
“Di-Lepton
Channel”
=> Signature-Based
W
 l,
qq,,W
W
lqq ~5%
~45%
l
~30%
W
10
Measuring Top Pair Production
(tt) 
Nevents - Nbackground
Luminosity
*

Production Cross Section
Why is measuring the rate of top production important?
• Higher cross section than predicted could be a sign of
One of
the first
things to measure
is the
non-standard
model
production
mechanisms
top pair production rate.
Resonant state X tt OR Anomalous couplings in QCD?
• It could also mean new physics in the top sample!
11
Finding Top Is Difficult!
In Run 1, we likely produced ~500 top quark pairs
at CDF. The problem was finding them.
We had only 76 ttbar pairs in our mass sample!
Separating Top from background:
•Finding Challenges:
clean lepton samples
•Tagging b-jets
Traditional
CDF method
–Displaced
vertices
Acceptance
for Top
–Soft
lepton tagging
(SLT)
(improved
in Run
2)
–Jet Probabilities
HT Distribution for Top
Events vs Background
Top Events
from
•Separating
Fitting to kinematical
distributions
using likelihood
or neural
network
W+jets
and QCD
Backgrounds
techniques
12
Top Cross Section Measurements
(Scorecard)
Top
AllProbability
Hadronic:
Kinematics:
B-tags
+
kinematics
Jet
Two
high
PNetwork
T Leptons
B-tags
Neural
+ kinematics
Separating Top from Background
Silicon b-Tagging
Soft Lepton Tagging
•Identifying clean lepton samples Charged
Particles
•Tagging b-jets
L
–Displaced vertices
New Results!!
-1
–Soft lepton tagging (SLT)
with > 300 pb
Secondary
Vertex
Impact
–Jet Probabilities
Primary
xy
Vertex
•Fitting to kinematical distributions
using likelihood or neural network
techniques
Parameter
162 pb-1-1Lepton+Jets
-1
-1 Lepton+Jets
194
pb
195
200
pb
DiLepton
-1 Lepton+Jets
PRD
71,
052003
(2005)
-1
162
pb
Lepton+Jets
162
pb
Lepton+Jets
Result
Submitted
to
PRD
-1 Allto
PRL
Submitted
93, 142001
(2004)
PRD
162
pb
Jets
Conference
Result
PRD 71, 052003 (2004)
Conference Result
13
Collected Dataset for CDF
Luminosity collected up until Fall ‘04 shutdown.
14
(Most early 2005 results: 318-347 pb-1)
Multivariate L+J Cross Section:
Neural Network
Previous Result: 195 pb-1 (PRD) . This Result: 347 pb-1
Total Event
Energy HT
|| Max Jets
Method: Use 7 kinematic and event shape variables to
discriminate top from background by training a neural
15
network to distinguish events.
Kinematics to Find Top
Background trained
on AlpGen+Herwig
W+3 parton MC
Signal trained
on Pythia ttbar
(tt)=6.0 ±0.8 ±1.0 pb
Adding
moreoutput
event
information
allows
Output
of
a
7-Input
neural
network,
choosing
both
Fit
to
neural
network
for
top
and
W+jets
background:
Neural network output shape templates for signal,
electroweak,
better
discrimination
ofdiscriminate
topbut
events.
shape
and
energy
variables
to
top
from
bkg
Sensitivity
similar
to
b-tagged
larger
used
16
and
QCD
multijet
backgrounds,
normalized
to unitsample
area.
Looks
like analysis,
top!
Kinematic Cross Section Results
Sample
Events
Fitted tt
(tt )
W +  3 jets
936
148.2  20.6
6.0  0.8  1.0 pb
W +  4-Jet
210
80.9  15.0
6.1  1.1  1.4 pb
In good agreement with theory for Mtop = 178 GeV:
(tt)=6.1 ± 0.8 pb (M. Cacciari, et al. JHEP 404, 68 (2004))
CDF Preliminary
(347 pb-1)
17
Keys to Improvement
 3 Jets
Effect
Jet Et Scale
Acceptance
3.0%
Shape
 4 Jets
Total
5.4%
8.3%
10.2%
QCD fraction
Acceptance Shape Total
3.3%
11.8%
10.2%
16.0%
16.0%
0.9%
0.9%
1.8%
1.8%
QCD shape
1.0%
1.0%
2.5%
2.5%
Other EWK
1.0%
1.0%
1.1%
1.1%
W+jets Q^2 Scale
8.6%
ttbar PDF
1.5%
2.9%
4.4%
2.4%
2.3%
4.7%
ttbar ISR
0.4%
1.2%
1.6%
2.0%
0.6%
2.6%
ttbar FSR
0.8%
0.7%
1.5%
0.7%
0.2%
0.8%
ttbar generator
1.7%
0.9%
2.6%
4.4%
0.2%
4.5%
Lepton ID/trigger
1.3%
1.3%
1.3%
1.3%
Lepton Isolation
5.0%
5.0%
5.0%
5.0%
Luminosity
Total
5.8%
5.8%
16.4%
18
22.8%
SecVtx B-tagging in Lepton+Jets
New tight SecVtx b-tagger:
•Tracking improved in new offline, cuts loosened
•Tightened secondary vertex quality requirements
•15-20% tag efficiency increase from previous tagger
Event tagging efficiency for ttbar:
19
Optimized L+J Cross Section Analysis
Improve Signal, S/sqrt(S+B), B:
•Signal: dataset doubled (162 318 pb-1), tagger improved
•S/(S+B): Re-optimize cut on HT as in previous analysis
•Background Error: Reduce error on poorly-modeled QCD
fakes by cutting out a lot of these backgrounds: MT(W) cut
HT> 200 GeV Optimal
MT(W) > 20 GeV Optimal
20
SecVtx B-tagged Cross Section
Backgrounds estimated from data and MC, the traditional CDF
Method 2. Top is excess above these for ≥3 jets.
Sample
Events
tt Fraction
(tt )
 1 b tag
138
81%
7.9  0.9  0.9 pb
 2 b tags
33
90%
8.7  1.7  1.5 pb
21
SecVtx Cross Section Systematics
Background Errors Reduced with Optimization
Largest systematic remains the b-tagging scale factor (6.6%). The
Heavy Flavor fraction, ~50% of background, ~2-3% on cross section.
22
SecVtx Double-Tagged Event
The new tagger has provided a clean sample of single- and
double-btagged events, which will be useful for single top, top
23
properties, and searches such as for WH.
CDF Cross Section Results Summary
Latest results in Red
Many different approaches to
measuring the top cross section,
allowing us to carefully
cross check the results-- and
look for anomalies.
24
Run 1: Excess in the b-tagged
Lepton + Jets Sample?
Observed excess of
b-tags in the 2 jet bin
Too many SVX double
tags (more than one btagged jet/event)
Too many multiple tags
(more than one b-tag/jet)
A lot of speculation,
but nothing solid.
25
Understanding Wbb+Jets
Use Data to Measure :
 (W + bb)
 (W +1,2 jets)
•Provides cross check on heavy
flavor fraction
•First step towards measuring Wbb
cross section
•Will help top properties
measurements, and searches for
single top, and for Higgs
Assume efficiencies cancel:
General Method:
•Create MC templates for the vertex mass of b, c, light quark jets
•Combine tagged MC events, and fit vertex mass distribution from
data
26
•Use the pre-tag sample for W+1,2 jets
W+bb/W+jj Ratio Results
Vertex Mass Templates from
data and monte carlo
Fit to b-tagged data, obtain
number of observed tags
Use SecVtx backgrounds, pretag estimates from data
Result:
 (W + bb)
 (W +1,2 jets)
 0.0072  0.0024  0.0022
SecVtxwill
HF improve
Fraction with
Prediction:
0.012±0.003,
1.3higher
Result
more data
and anti-charm
tagging27
Top Mass: Current World’s Best
Details: Wine & Cheese April 12th
By U.K. Yang
Key Ingredients:
•Statistics: Large sample of
single and double tagged
events
•Systematics: Simultaneous fit
to top mass and jet energy
scale using Wjj decays
M top  173.533..76 (stat. JES) 1.7 (syst.)GeV/ c2
28
New Top Mass:
Updated Dynamic Likelihood Result
Previously, the best Run 2 result with 162 pb-1
Integrate
over z1, z2,
parton pT
Sum over
all possible
jet-parton
assignments
Parton
Distribution
Functions
+ ISR
Replace pz with
W propagator
factor
Matrix element
provides complete
dynamical event
information
Transfer
functions
connect
jets to
partons
2 4
L (M top )   
F ( za , zb , pT ) | M |2  (sw  (  ) 2 )w( I t , x | y; M top )dxdsw
Flux
It I s
i
Likelihood for Each Event i
29
New DLM Top Mass Result
No ME used for backgrounds.
Instead, mapping function used
since backgrounds dilute mass
in a known manner
Mtop = 173.8 ± 4.2
30
DLM Top Mass Systematics
Sources
Mtop(GeV/c2)
JES (up to hadron)
2.1
T.F.(up to parton)
2.2
ISR
0.4
FSR
0.5
PDF
0.5
Generator
0.3
Run I
bkg fraction (6.7%)
0.6
Run II 2004
bkg Modeling
0.6
b tagging
0.2
b jet energy
0.6
Total
3.3
Fractional Syst. Uncertainty vs PT
By far the largest improvement has
been the reduction in the jet energy
scale systematic! 5.3 GeV3.0 GeV
Central  region
~3% jet PT
uncertainty in
top events
31
New Top Mass Comparisons
Method
163 pb-1 result
Template
176.7 ± 9.1 (“1D”)
DLM
177.8 ± 7.8
318 pb-1 result
Selection
Combinatorics
JES
173.5 ± 4.1 (“2D”)
173.8 ± 4.2
≥ 4 jets
Best 2
Wjj
= 4 jets, ≥ 1 tag
Use all
None yet
Background
Template
Mapping
• Complementary methods
–Different sensitivity to details of production
and decay.
• DLM could in principle implement similar
treatment of JES in the joint likelihood.
32
CDF Top Mass Summary
Two new top mass results:
Best is more sensitive than the
Run 1 combined!
On Deck:
Channel
L+Jets Dilepton All-had
Analyses
6
5
2
Mature
2
4
0
Blessed
2
0
0
Renewed Tevatron Top Mass
Combo Meeting held June 9th…
Stay tuned!
Publications and combination
Coming soon!
33
Future for Top Mass
Our dominant systematic, the jet energy scale, now
scales with statistics!
34
Top Decay Properties
We said tWb, but really 100%?
Indirect measurement using the CKM matrix:
• Elements |Vub| and |Vcb| measured to be very
small from decay of B mesons
• Unitarity and only three generations implies
|Vtb| is 0.998 @ 90% CL
With top quark samples we can measure it directly as “R”:
R
2
BR(t  Wb )

BR(t  Wq ) Vtd 2  Vts 2  Vtb 2
Vtb
where q  {d, s, b}
Use the ability to identify jets with a distinguished secondary vertex: b-tagging
•The number of b-tagged jets depends strongly on R and eb
We classify the ttbar sample based on the number of b-tagged jets
•The relative rates of events with 0/1/2 b-tags is very sensitive to R
35
Measuring BR(tWb)/BR(tWq)
•Use the Lepton+Jets and Dilepton samples.
•Total integrated luminosity of 162 pb-1
•Classify samples having 0/1/2 b-tagged jets
•Estimate background contribution to each of
the six sub-samples
•MC and data driven (Method 2)
•Background Lepton+Jets 0-tags obtained
using NN techniques.
•determine the b- c- and q- jet tagging
efficiencies εb, εc and εq; then find efficiency
to have 0,1 or 2 tags in a particular top event.
•get the expected top content in 0/1/2 tags.
36
Measuring BR(tWb)/BR(tWq)
Compare the expected top with the observed top in the 0/1/2
tag subsets and extract R by maximizing the likelihood.
0.17
R  1.1200..21
190.13 (stat  syst)
Set F-C lower limit :
R >0.61 at 95%CL
|Vtb| > 0.79 at 95%CL
(assuming unitarity)
Mild excess in double b-tags sample drives the R value above 1 37
”R” Consistent with Standard Model
So…….
This means that the top decays to a b
quark most of the time, as expected.
t
?
But, is
Could
?
?
b
always a W+ ?
be sometimes an H+ ?
38
Measurement of BR(tH±b)
•
assume each top quark has 5 possible decay modes
t  Wb
t  H+b  t*bb  W+bbb
t  H+b  b
t  H+b  csb
t  H+b  W+h0b  W+bbb
•
•
Data: use four XS samples
–
–
–
–
dilepton
lepton+jets (1 tag)
lepton+jets (2 or more tags)
lepton+h
exp
back
N XSA
 N XSA
   tt , XSA
from XS meas.
theory=(6.7±0.7)pb
(hep-ph 0303085)
(CPsuperH, full SUSY EW/QCD corrections)
 Ldt
~191 pb-1
from MC
 tt , XSA   Bi B j  i , j XSA wTop, wHiggs, mH , mh
5

i , j 1
Branching fractions
of each decay mode
39
0

Expected Events v. tan(b) Per Sample
Integrated Luminosity of 171 pb-1
40
Limits: MH+ v. tanb, Min Stop Scenario
BR’s predicted by MSSM in Minimal Stop Mixing scenario
Typical search for h0 at LEP(hep-ph/9912223).
41
What Can We Take from This?
There is no evidence within reach for top decaying
to charged Higgs.
So…….
Assume that top decays to W+b.
t
W+
b
 But, is the nature of the tWb vertex as expected?
42
W Helicity from tWb Decays
•
Examines the nature of the tWb vertex,
probing the structure of weak
interactions at energy scales near EWSB
•
Stringent test of SM
and its V-A type of interaction.
V-A Suppressed
W0 Longitudinal fraction W- Left-Handed fraction W Right-Handed fraction
+
FF0
F
+
0
+1/2
t
W
W
-1/2
+1/2
t
b
t
W
+1/2
+1
+1/2
W
b
+1
-1/2
43
Run 2 W Helicity in Top Events
SM: Only longitudinal and
left-handed W’s can be
produced in the top rest
frame.
Use lepton pT spectra to
determine the fraction
F0 of longitudinally
polarized W’s.
F0 = 0.7 in the Standard Model
The combined dilepton and
lepton+jets b-tagged events
plotted against the best fit.
Result: F0 = 0.27+ 0.35 – 0.21
Or
F0 < 0.88 @ 95% CL
44
Run 2 W Helicity Using cos*
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
Both dilepton and lepton+jets
events combined. Combination
with pT & F+ underway…
Result: F0 = 0.89 ± 0.32 ± 0.17
Or
F0 > 0.25 @ 95% CL
45
What About Production?
• We know that, within errors:
– The top decays mostly to W+b
– The nature of the tWb vertex is what’s expected.
• Measured Production Cross Sections Consistent
with Standard Model, within errors
W+
t
X
b
W+
t
b
 Are some of the top-like events from a heavy top?
 Are some small number of top pairs coming from a resonance?
46
Search for High ET Top-Like Events
HT distribution for W+4p, ttbar, and t' where M(t')=225 GeV
We can set limits on new physics processes in top sample
47
HT Plot with t' Signal, M(t') =225 GeV
Plot for fit result with t' signal included at 95% CL limit
48
[(ttbar) 6.12 pb in this fit]
Result: Limits (pb) Versus M(t')
Constraints
vary with
assumed
top mass, but
not by much.
Mtop
constraint *
170
7.8  1.0 pb
175
6.7  0.9 pb
180
5.75  0.7 pb
* Taken from
hep-ph/0303085
Mtop=180
Mtop=170
Expected
1 sensitivity
Limits on BR(t'Wq)2
49
Projected Limits: Higher Luminosity
Large improvement in systematic errors expected:
(Jet energy scale dominates!) Should do better! 50
Run 1 Searches for ttbar Resonances
Investigate models that
dynamically break EW
symmetry, such as topcolorassisted technicolor
Coming in Summer ‘05:
Exclude a narrow, leptophobic
X boson with mX < 480 GeV/c2
Search for narrow
model-independent Xtt
resonances in l+jets
Run 2 CDF Matrix Element
Analysis Xttbar
51
What Else is in the Top Sample?
• We see what looks like top so far, but with more
statistics, we can probe kinematics and
properties further, to see if there are any non-SM
events
~
t

b
W+
t
b
 Are some of the top-like events from SUSY or other new physics?
 Are top kinematics as we expect?
52
Run 1: Anomalies in the Top
Di-lepton Sample?
Four di-lepton
candidates with
very high MET.
?
53
Run 2: Di-lepton Kinematics
54
Distributed as expected? More data to come…
PT(lepton)
Analysis of SM Agreement
Probability Using Kinematics
 = 1.6%
Topological weighting parameter (TW)
Probability() of CDF outcome with SM hypothesis
55
using 4 kinematic variables [MET, Tw, PT, (MET,l )]
Kinematic Discriminants Look Standard
MET
pT lepton

T
Result driven by low pT excess. Consistent with SM hypothesis.
56
More data will allow better sensitivity of such tests in the future.
CDF Top Physics Publications
Top Pair Production Cross Section
Measurement of the ttbar Production Cross Section in ppbar Collisions at sqrt(s) = 1.96 TeV using Dilepton Events
Submitted 04/27/04, accepted August 2004 Phys. Rev. Lett. 93, 142001 (2004)
Measurement of the ttbar Production Cross Section in ppbar Collisions at sqrt(s) = 1.96 TeV using Lepton + Jets Events
with Secondary Vertex b-tagging Submitted 10/14/04, accepted March 2005 Phys. Rev. D71, 052003 (2005)
Measurement of the ttbar Production Cross Section in ppbar Collisions at sqrt(s) = 1.96 TeV using Kinematic
Fitting of b-tagged Lepton + Jet Events Submitted 09/09/04, accepted April 2005 Phys. Rev. D71, 072005 (2005)
Measurement of the ttbar Production Cross Section in ppbar Collisions using the Kinematics of Lepton+Jet Events
Submitted 04/27/05 to Phys. Rev. D hep-ex/0504053
Measurement of the tt-bar Production Cross Section in ppbar Collisions at sqrt(s) = 1.96 TeV Using Lepton Plus Jets
Events with Semileptonic B Decays to Muons Submitted 06/01/05 to Phys. Rev. D hep-ex/0506001
Search for Single Top Production
Search for electroweak single top quark production in ppbar collisions at sqrt(s)=1.96 TeV
Submitted 10/20/04, accepted January 2005 ･Phys. Rev. D, 71 012005 (2005)
Top Properties
Search for Anomalous Kinematics in ttbar Dilepton Events at CDF II
Submitted to Phys. Rev. Lett. 12/10/04, accepted June 2005. hep-ex/0412042
Measurement of B(t --> Wb)/B(t --> Wq) at the Collider Detector at Fermilab
Submitted to Phys. Rev. Lett. 05/27/05 FERMILAB-PUB-05-219-E.
57
Near Term Plans
Summer conference results will include
•new top mass measurement in the dilepton channel
•matrix element analysis of X ttbar
•Tau + jets ttbar cross section
•Update on all hadronic cross section
•Update on dilepton cross section…
CDF will shoot for presenting physics results on 1 fb-1 for
Winter Conferences 2006!
58
Summary
Many more analyses are testing top properties, and
hunting for hints of new physics in the top quark datasets:
•SecVtx and SLT combined tags in W+jets
•Top spin, top charge, top lifetime
•Ongoing search for single top in lepton + jets events
•Tests of consistency of kinematics with the SM in the
top quark sample
•Rare top decays (tZc) and FCNCs
•Top production (q-qbar v. glue-glue)
59
Conclusions
As we increase our datasets at the Tevatron in Run 2,
CDF will have much to say about the top quark, it’s
properties, and the possibility of
new physics in our top quark samples.
Stay tuned:
CDF top quark publications are rolling…
60
Searches for Single Top
Combined Channel Search:
(s+t) < 13.7 pb @ 95% C.L.
t-Channel Search:
(t-chan) < 8.5 pb @ 95% C.L.
We are still looking for single top!
61
Standard Model Higgs?
top and W masses
constrain the mass of the
Standard Model Higgs
LEP Direct Search Limit:
Mass (Higgs) > ~114 GeV
(World Data 2 fit: MH ~ 96 GeV)
62
Can a t' Exist?
•Z width measurement rules out a fourth generation with a
light neutrino m(4)<m(Z)/2
•He/Polonsky/Su (hep-ph/0102144): a generic 4th chiral
generation is consistent with EWK data; can accommodate
a heavy Higgs (500 GeV) without any other new physics
(similarly with 2HDM)
•Some Little Higgs theories predict a heavy top T at or
below the TeV scale (reference)
•N=2 SUSY requires three more “mirror” generations – the
SUSY breaking mechanism can induce couplings of the
mirror quarks with the known ones
•Other models (eg: “Beautiful Mirrors” hep-ph/ 0109097)
include possibilities of a new heavy up-type quark
decaying to Wb
63
Neural Network Details
Neural Network Training:
Seven input variables
HT
Aplanarity
Maximum jet 
ET(Jets 3,4, and 5)
NN structure
 pZ/ ET
 Minimum dijet invariant mass
 Minimum dijet separation (R)
Seven input nodes
One hidden layer with seven nodes
One output node (signal target = 1, background = 0)
Feed-forward network
Trained using Root_JetNet
64