Top quark production cross-section measurements at D

1
Top quark production cross-section measurements
at DØ

• Flera Rizatdinova (KSU)
• for the DØ collaboration

• Introduction
• Leptonjets topological analysis (see WC talk
  by B.Abbott)
• Leptonjets analysis based on b-tagging
• Conclusions
• Outlook

2
Motivation for the top quark studies

• Top quark has been discovered by CDF and DØ
  collaborations in 1995 with data of 50 pb-1
• Top quark is the only known fermion with a mass
  on the electroweak scale
• Study of the top quark provides an excellent
  probe of the electroweak symmetry breaking
  mechanism
• New physics may be discovered in either its
  production or decays
• Tevatron is the only place to study top quark
  properties before LHC operation.

3
Top quark production and decay

• Br(t ? Wb) ?100 in the SM

• in proton-antiproton collisions
• at Tevatron energies (?s 1.96 TeV), top quarks
  are mostly produced in pairs

Both Ws decay via W?l? (le or ? 5)
dilepton One W decays via W?l? (le or ? 30)
leptonjets Both Ws decay via W?qq (44) all
hadronic
90
10
EW single top production not yet observed
4
DØ detector
5
Silicon Microstrip Detector (SMT)
B?J/?K
SMT combines vertex and tracking capabilities and
provides good primary and secondary vertex
resolutions.
6
Lepton jets channels

• Golden mode for top studies 30 yield and
  relatively clean

Event preselection
jet

• 1 high pT isolated charged lepton (e,m).
• Neutrinos large missing ET
• Large jet multiplicity
• dilepton veto

?
Backgrounds
p
b

• Wjets and fake leptons in QCD

jet
Further selection techniques
jet

• topological analysis n ? 4 jets
• tag b jets with Soft Lepton Tag (SLT)
• ? 3 jets, ? 1 SLT
• tag b jets with displaced VTX or IP
• ? 3 jets, ? 1 b tag

jet
t(?Wb) t(?Wb)
e,m
qq
7
Leptonjets topological analysis

• Event selection
• Preselect a sample enriched in W events (leptons
  with pTgt20 GeV, ETgt20 GeV)
• QCD background estimation (from the data)
• Separate W and QCD events with loose (L) and
  tight (T) lepton characteristics.
• Difference between loose and tight samples for
  ejets is in the track match to EM object for
  mjets is in the muon isolation.
• Matrix method is a way to calculate fractions of
  signal and BG based on different probabilities
  for a high pT lepton from signal eWtt and from
  QCD BG eQCD to pass selection criteria
• W4 jet BG estimated from the data with Berends
  scaling.

?
8
Leptonjets topological cuts and SLT
ejets (49.5 pb-1) mjets (40 pb-1)
ljets with SLT
ET(W) gt 60 GeV MET(Cal) gt 15 GeV ET(W)/ MET(Cal)
hW lt 2 PT(leading jet) gt55GeV hW or PT(leading jet)
HT gt 180 GeV HTall gt 220 GeV HTgt110 GeV
Aplanarity gt 0.065 Aplanarity gt 0.065 Aplanarity gt 0.04
e
NW NQCD All BG Exp tt Nobs
ejets 1.3?0.5 1.4?0.4 2.7?0.6 1.8 4
mjets 2.1?0.9 0.6?0.4 2.7?1.1 2.4 4
ejets(SLT) 0.0?0.1 0.2?0.1 0.2?0.1 0.5 2
mjets(SLT) 0.2?0.1 0.4?0.1 0.6?0.1 0.4 0
s( ) 7pb
9
Cross-section from topological analyses
DØ Run II preliminary
leptonjets channels only
all combined
10
Leptonjets with b-tagging Method overview
data only
Estimate production cross-section from the
excess observed in the number of tagged events
w.r.t. BG expectation in 3 and 4jet multiplicity
bins.
11
Two b-tagging methods
Counting Signed Impact Parameter tag (CSIP)

• S IP/?(IP)
• Jet is positively tagged if it has
• at least two tracks with Sgt3 or
• at least three tracks with Sgt2

• Look for displaced vertices ( 2 tracks),
• jet is tagged as b jet
• If signed decay length significance gt5

Secondary Vertex Tag (SVT)
12
Signal data

• The ejets channel L 49.5 pb-1

N W1j
W2j W3j
W4j preselected data 2599
741 146
25 Wej 200454
46230 7115
124 QCDej 5959
2798 756
133

• The mjets channelL 40.0 pb-1

N W1j
W2j W3j
W4j preselected data 2796
973 217
40 Wmj 2188100
65054 12623
259 QCDmj 60885
32344 9120 136
13
Primary vertex selection

• zPV 60 cm
• N of tracks in the PV 3

• Efficiency measured on
• EMqcd sample requiring signal trigger for
  ejets channel
• mjets loose sample for mjets channel

Efficiencies for PV cuts
1j 2j
3j 4j ejets
92.270.06 93.80.1 95.00.3
95.20.4 mjets 89.00.3 90.80.3
91.70.4 91.80.7
14
Jet tagging probability

• The probability to tag a jet was split into two
  components
• the probability for a jet to be taggable (at
  least 2 tracks with pTgt0.5 GeV and c2lt3 in the
  jet cone DRlt0.5 with at least 3 hits in silicon
  detector (SMT) or 2 SMT hits in the innermost SMT
  layers) - TAGGABILITY
• the probability for a taggable jet to be
  tagged - TAGGING EFFICIENCY

15
Taggability
0.8

• Taggability was measured on the signal sample
• The fit to the data distributions vs ET and h was
  performed
• Two-dimensional parameterization was obtained
• Cross-checked the 2-d parameterization applying
  it back to the data.

0.4
0.0
0.0
1.0 2.0
-2.0 -1.0
0.8
0.4
20 40 60 80 100 ET
0.8
0.4
0.0
Data Predicted by fit
h
100 50 0
ET,GeV
-2 0 2
50
100
-2 -1 0 1 2 h
16
B-tagging efficiency
SVT CSIP
B-tagging efficiency was measured by three
different methods and compared to the MC
prediction. All measurements are in a good
agreement with each other. Average b-tagging
efficiency for basic method
0.4
0.2
20 40 60 80 100 ET
0.3
SVT CSIP
?btag (31.91.6) (36.31.9)
0.1
0 1 2 h
17
B-tagging efficiency calibration
We use MC to calculate Wjets BG and top
expectation Þ have to use b-tagging efficiency
calibrated to the data. The way to calibrate -
introduce SFbm(ET, h)
Shown two-dimensional scale factor SF are
obtained as a products of one-dimensional scale
factors assuming that they are not correlated.
SVT
CSIP
18
c-tagging efficiency
c-tagging efficiency was not measured directly on
the data. Instead, we use b-tagging efficiency
measured on the data and apply to it the
correction factor CFb?c. Correction factor
CFb?c is calculated using Monte Carlo. Resulting
c-tagging efficiency is
19
Negative tagging rate
Measured on data
0.8 0.6 0.4 0.2
2 1
CSIP SVT
1.2 0.8 0.4
-2 -1 0 1 2
20 40 60 80 100
Jet pT
Jet h
-2 -1 0 1 2
Jet h
Negative tagging means SVT signed decay length
of SV lt-5 CSIP signed IP of tracks is lt-3 for
two tracks or lt-2 for three tracks in jet
2 1
Jet pT
20 40 60 80 100
20
Estimation of the mistagging rate

• Need to calculate the contribution from Wlight
  jets events to the total BG.
• Measured negative tagging rate on data, but want
  to know the probability to tag a light jet (jet
  originated from u,d,s quarks).
• Need to correct negative tagging rate
• for the presence of heavy flavor in data in
  negative tags (correction factor SFhf)
• for the missing contribution from long-lived
  particles (correction factor SFll)

From MC studies
SVT CSIP
1.110.11 1.090.09
21
MC samples for Wjets BG estimation

• Wjets with ALPGEN 1.1

We generated 14 processes with W and various
numbers of partons of different flavors. From
these samples we determine fractions of the
different Wjets flavor processes contributing to
each exclusive jet multiplicity bin.
22
Wjets background estimation

• Estimated number of tagged Wjets events
• Particular case Expected contribution of Wnj
  events after SVT tagging in mjets channel is
  shown in the table

N of preselected Wnj events before tagging
Average event tagging probability
Source W1j W2j W3j W4j
Wlight 4.711.22 2.810.71 0.770.26 0.230.10
Wbb 0.000.00 1.520.29 0.520.15 0.260.13
Wcc 0.000.00 0.770.23 0.290.11 0.110.06
W(bb) 1.490.75 1.901.15 0.270.18 0.040.05
W(cc) 0.980.47 0.680.41 0.350.16 0.130.10
Wc 2.850.86 2.310.80 0.260.14 0.090.07
Expect 3.34 events from Wnj BG after tagging in
W3j and W4j event topologies
23
QCD background in ejets channel
Sources of QCD BG - fake Compton QCD and fake
electrons (jets) Reminder NQCD events before
tagging is estimated using Matrix
Method Measured probability to tag a QCD event
PQCDon data directly for the different jet
multiplicity bins NQCD after tagging is product
of NQCD and PQCD
PQCD W1j W2j
W3j W4j SVT
(1.020.03) (1.810.08) (2.710.26)
(3.300.73) CSIP (1.260.03)
(2.190.09) (3.330.28) (3.800.77)
Numbers of predicted QCD events after tagging
NQCD W1j W2j
W3j W4j SVT
2.980.15 3.450.23 1.620.23
0.390.14 CSIP 3.690.17
4.190.28 1.980.27 0.450.16
24
QCD background in mjets channel
Sources of QCD BG - semileptonic heavy flavor
decays. Could not use the same method as for
ejets since the anti-W cuts used to select pure
mQCD sample affect the flavor composition. Final
method apply the Matrix Method to the tagged
signal sample and obtain NQCD in the tagged
sample directly.
Numbers of predicted QCD events after tagging
NQCD W1j W2j
W3j W4j SVT
3.901.50 3.001.10 1.900.70
0.700.30 CSIP 4.591.40
3.441.22 1.940.83 0.610.41
25
Expected tt yield in ejets channel
N of expected tt events after tagging
Ptagtt is a probability to tag a top event -
calculated in the same way as probability to tag
a Wnj event esel and etrig selection and
trigger efficiencies for tt events ePV -measured
on data for both ejets and mjets channels
Ptt W1j W2j
W3j W4j SVT
(16.693.30) (27.883.41)
(36.834.14) (42.354.66) CSIP
(18.422.94) (30.263.56) (39.604.34)
(45.754.86)
Numbers of expected tt events after tagging
(assuming s(tt) 7 pb)
Ntt W1j W2j
W3j W4j SVT
0.020.01 0.400.06 1.350.18
1.680.26 CSIP 0.020.01
0.430.07 1.450.19 1.810.28
26
Expected tt yield in mjets channel
Ptt W1j
W2j W3j W4j SVT
(12.502.91) (27.373.45)
(36.154.13) (41.334.61) CSIP
(15.732.39) (30.273.59) (39.614.36)
(45.564.84)
Numbers of expected tt events after tagging
(assuming s(tt) 7 pb)
Ntt W1j W2j
W3j W4j SVT
0.010.01 0.240.04 1.200.17
2.000.31 CSIP 0.010.01
0.260.04 1.310.18 2.200.34
27
Systematic uncertainties

• Took into account 26 different sources of
  systematic errors
• Largest uncertainties
• Jet Energy Scale (1.25 pb)
• NW and NQCD in data (0.8 pb)
• Tagging probability (0.8 pb)
• Semileptonic b-tagging efficiency in MC (0.5-0.8
  pb)
• Semileptonic b-tagging efficiency in data (0.5
  pb)
• W fractions from gluon splitting in HERWIG (0.5
  pb)

• Preselection efficiency
• Trigger efficiency
• PV selection efficiency
• NW and NQCD in data
• Tagging probability
• W fractions from ALPGEN
• Track matching with EM cluster
• Electron identification efficiency
• Muon identification efficiency
• Jet identification efficiency
• Jet resolution
• Jet Energy Scale
• Taggability in data
• Flavor dependence of taggability
• Inclusive b-tagging efficiency in MC
• Inclusive c-tagging efficiency in MC
• Semileptonic b-tagging efficiency in MC
• Semileptonic b-tagging efficiency in data
• Negative tagging rate in data

Selections
Object ID
Tagging probability
MC
28
Observed numbers of events vs predicted
Combined lepton jets channels
D0 Run II preliminary
D0 Run II preliminary
CSIP
lepton jets 1 jet 2 jet 3 jet 4 jet
Before tag 3681 1351 298 65
W jets 22.3 4.7 18.7 3.4 4.4 0.9 1.4 0.4
QCD 8.2 1.4 7.6 1.2 3.9 0.9 1.1 0.4
Total BG 30.6 5.0 26.4 3.5 8.3 1.3 2.5 0.7
Expected 0.70.1 2.8 0.2 4.0 0.6
BG 30.6 5.0 27.1 3.6 11.1 1.4 6.5 1.0
tagged 34 27 13 6
Good agreement with BG expectation in the first
two bins (no top contribution is expected there),
clear excess observed in the number of tagged
events over predicted BG in third and fourth jet
multiplicity bins.
29
Comparison of b-tagging results with topological
analysis
e jets
µ jets
SVT
Many tagged events in ejets channel pass
topological cuts
DØ Run II preliminary
Combined ljets channels N of tagged events and
BG composition
30
tt production cross-section
Look at the excess of tagged events over
predicted ones in 3jet and 4jet multiplicity
bins Perform a maximum likelihood fit to the
observed numbers of events
of tags W1 jet W2 jet W3 jet W 4 jet
CSIP 34 27 13 6
SVT 28 20 9 9
L 45 pb-1 (s 7 pb)
CSIP
SVT
31
Tagged Event
Event is tagged by both algorithms (run 169923
event 16396718) Njets 4 pT(e) 27 GeV , MET
58 GeV pT(jet) 51, 36, 30, 53 GeV HT 207
GeV Aplanarity 0.11
Primary vertex Ntrack 17 z ?4.6 cm
32
Conclusions

• DØ has re-established top quark signal in the
  majority of top quark decay modes
• DØ presents its first results on measurement of
  the top quark cross-section with lifetime
  b-tagging
• Results are obtained on Moriond data samples with
  limited statistics and controlled systematics
  errors
• Expect to have much better results by Winter
  Conferences

Topological
SVT CSIP
33
Outlook

• The DØ Run II measurements of the top pair
  production cross section demonstrate significant
  progress in the optimization and understanding of
  the detector performance
• There is a big potential to improve crucial
  aspects of physics analyses (tracking in jets,
  physics object identification, b-tagging
  optimization and many others)
• With larger statistics and better understanding
  of the data by the end of this year we expect to
  perform top quark measurements with greatly
  improved precision

34
Background slides
35
Leptonjets topological analysis
Berends scaling
DØ Run II Preliminary
a 0.145?0.02
a 0.164?0.02
Estimation of the background for Njets? 4 (from
data, using matrix method)
24.2
11.9
22 (ejets)
38 (mjets)
11.9
12.5
36
Leptonjets Analysis with soft lepton tagging
Selection before Soft Muon Tag

(Loose/Tight sample)
- Use the same preselection as ljets - Require
at least 3 jets - Apply mild topological
cuts (HTgt110 GeV, Aplanaritygt 0.04)
75/23 (mjets)
?
459/27 (ejets)
DØ Run II Preliminary
QCD background from matrix method
W bkg. from Tag rate functions
0.4?0.1 (m)
0.0?0.1 (e)
Analysis Bkg. Tot. Sig. Nobs
ejets 0.2?0.1 0.5 2
mjets 0.6?0.3 0.4 0
For s 7pb
37
Cross-check of the mistagging rate
Know light jet tagging efficiency, b and
c-tagging efficiencies in data can predict
positive tagging rate in data which includes
light and heavy flavor jets
here all efficiencies and fractions are functions
of (ET,h) and
(Fb and Fc were obtained from QCD MC)
Good agreement between observed and predicted
tagging rates!
38
Scale Factor (CSIP)