1. Tim Scanlon Imperial College, London on behalf of the DØ Collaboration
Search for SM Higgs Boson Using Large Missing Transverse Energy and B-jets at DØ
Tim Scanlon
Imperial College, London
on behalf of the
DØ Collaboration
Overview:
Introduction
The DØ Detector
Previous Results
Analysis Method
Published Result
Future Analysis
Conclusion

2. Introduction Z Motivation
ZHbb is a very sensitive way to search for the SM Higgs at the Tevatron as we do not distinguish between the neutrino species
(qqZH)xBr(Z, Hbb) = mH=115 GeV
(qqWH)xBr(Wl, Hbb) = pb
Characteristic Signal
Large missing ET (ET)
2 b-tagged jets with high pT
The leading jets are boosted
and hence not back-to-back
Di-jet mass of b-jets
No isolated leptons Z H n b
Jet1
Jet2 ET
Tim Scanlon (Imperial College London)

3. The DØ Detector
Good calorimeter (ET) and tracking (b-tagging) essential
Uranium/Liquid-Argon Calorimeter
Central calorimeter provides coverage up to ||~1.1
Two end calorimeters extend coverage up to ||~4.2
Tracking
Silicon Microstrip Tracker (SMT)
New Layer 0 for Run IIb
Central Fibre Tracker (CFT)
Surrounded by 2T Solenoid
DØ detector
Efficiency above 85%
Recorded 1.5 fb-1 of data
Tim Scanlon (Imperial College London)

4. Previous Preliminary Result
95% C.L. limits on (ppZH) × Br(Hbb) = 8.5 ~ 12.2
Updated version of the analysis now accepted for publication:
Same dataset
Improved:
Optimized event selection
Added “exclusive” single b-tag channel to double b-tag channel
Inclusion of WH limits in ET+jets sample, when the lepton from the W is missed
Tim Scanlon (Imperial College London)

5. Analysis Issues
The signal is well defined although it has significant backgrounds:
“Physics” backgrounds
Well defined processes can be distinguished and accurately modeled
Some irreducible
Dominant physics backgrounds
W+jets, Z+jets, top, ZZ, and WZ
“Instrumental” backgrounds
Basically everything else
Mainly QCD multi-jet events with mismeasured jets
back to back jets events where one jet is grossly mis-measured
Large ET
presence of fake jets, etc.
Generally low acceptance, but cross-section much larger
Significant background
No easy way to estimate the magnitude and shape of this background
Instrumental background forced us to apply more stringent selection cuts
Needed to devise a way of estimating and simulating its contribution
Tim Scanlon (Imperial College London)

6. Event Selection ET > 50 GeV (basic Higgs signal)
2 or 3 jets with pT > 20 GeV, ||<2.5 (basic Higgs signal)
(dijet) < 165° (rejects QCD di-jet events)
Isolated EM and muon veto (rejects top, W/Z+jets)
HT < 240 GeV (rejects top)
PTtrk > 20 GeV (rejects instrumental)
-0.1 < A(ET,HT) < 0.2 (rejects instrumental)
min (ET,jets) > 0.15 && ET > -40 * min (ET,jets) + 80
(ET, PTtrk) < 90° for “signal” region
> 90° for “sideband” region
Determined by optimisation.
- HT = |pT(jets)|
PTtrk = |pT(tracks)|
- A(ET,HT) = (ET-HT)/(ET+HT)
(Used to measure
instrumental background)
Variables verified in W+jets sample.
Tim Scanlon (Imperial College London)

7. Estimating the Instrumental Background
Tim Scanlon (Imperial College London)

8. Instrumental Background
Fit of A(ET, HT)
Estimate physics background from MC: Triple Gaussian function
Instrumental background: 6th order polynomial function
Instrumental background = ± 91.4 events (from fit)
Physics background = events (from MC)
Normalise sideband region instrumental background in A(ET, HT) bins
Model the instrumental background distributions in signal region
Signal Region
Sideband Region
Tim Scanlon (Imperial College London)

9. b-tagging Jet Lifetime Impact Parameter Tagger Identifying a b-jet
Track Impact Parameters (JLIP and CSIP)
Secondary Vertex (SVT)
High pT Lepton
Neural Network Combination (more later)
Jet Lifetime Impact Parameter Tagger
JLIP identifies heavy flavour jets from large impact parameter tracks
JLIP calculates a probability (P) that the jet is a light-jet
Analysis split into two different b-tagging channels
One JLIP tag ‘Exclusive’
Ultra Tight JLIP (P < 0.001)
Two (or more) JLIP tags
Loose JLIP (P < 0.01)
Extra Loose JLIP (P < 0.04)
JLIP Performance in Data
Tim Scanlon (Imperial College London)
(~P)

10. Distributions ET+jj Data : 3210 Exp : 3211 ET+jj (1 btag) Data : 592
ET+jj (2 btags)
Data : 25
Exp : 27.0
Tim Scanlon (Imperial College London)

11. Background Composition and Acceptance
Double (Single) Tagged Channel (within ±1.5 mass window)
ET+bb (bj)
(105 GeV)
(115 GeV)
(125 GeV)
(135 GeV)
# Data
10 (29)
11 (33)
10 (37)
9 (44)
# Predicted BKG
8.9 ± 1.7 (32.2 ± 5.9)
9.4 ± 1.8 (34.0 ± 6.1)
9.8 ± 1.8 (35.2 ± 6.0)
10.5 ± 2.0 (37.3 ± 6.6)
# ZH (Hbb)
Acceptance (%)
0.25 (0.24)
0.86 ± 0.16
0.21 (0.20)
1.04 ± 0.20
0.15 (0.14)
1.18 ± 0.22
0.091 (0.087)
1.34 ± 0.24
# WH (Hbb)
0.18 (0.18)
0.36 ± 0.07
0.43 ± 0.08
0.098 (0.096)
0.47 ± 0.09
0.062 (0.061)
0.55 ± 0.10
Large contribution from WH decays
Single Tag Channel
Main background is Wj
Instrumental background is 26%
Double Tag Channel
Main background top decay
Instrumental background reduced to 13%
On same level as the W+jj and Z+bb
Systematics: Signal 19%, Background 19%
b-tagging (~14%) and Jet energy scale (~8%)
Composition
Single
Tag (%)
Double Tag (%)
Zjj 8 3
Zbb 5 16
Wjj 38
Wbb 12
Top 33
WZ/ZZ 1 7
Instrumental 26 13
Tim Scanlon (Imperial College London)

12. (ppZH)xBr(Hbb) Limit
Expected/Observed Limits
105 Gev
115 Gev
125 Gev
135 Gev
ZH Limits (pb)
3.1/3.4
2.7/3.2
2.4/2.9
2.1/2.5
WH Limits (pb)
7.6/8.3
6.3/7.5
6.0/7.4
5.0/6.3
Significant progress since preliminary result
New limits are more than 2 times better
Limits set from combined double and exclusive single tag channels
Also measured limits for WH with escaped lepton
Results combined with other DØ and CDF result
Tim Scanlon (Imperial College London)

13. Significant progress made on next generation of analysis
Future Analysis
Significant progress made on next generation of analysis
Several improvements expected to significantly improve the limit
 1 fb-1 of data
Full calibration of calorimeter
Lower systematic errors
New NN b-tagging
NN event selection
New preliminary limit expected by early 2007
Tim Scanlon (Imperial College London)

14. New NN b-tagging (released summer 2006)
A new b-tagging tool
Combines various variables from the track based b-tagging tools in a Neural Network
Substantial improvement in performance over constituent input b-taggers
Trained on Monte Carlo
Certified on data
Performance measured on data
Increase of 1/3 in efficiency for a fixed fake rate of 0.5 %
Significantly increase Higgs sensitivity MC
Fake-jets with a very loose tag
Data
Tim Scanlon (Imperial College London)

15. Search for the Higgs boson with large missing transverse energy
Conclusions
Search for the Higgs boson with large missing transverse energy
Very important channel in search for the SM Higgs
Search for 2 b-tagged jets and large missing ET
Main difficulty predicting the instrumental background
0.3 fb-1 analysis accepted for publication in PRL
Results were two times better than preliminary result
Also measured WH with escaped lepton
1 fb-1 preliminary result expected early next year
Numerous improvements
Expect significantly improved limit
Current and future results
Vital role in combined SM Higgs search
Tim Scanlon (Imperial College London)

16. Backup Slides

17. Each systematic source varied by ±1
Systematic Errors
Signal Errors
Single Tag (%)
Double Tag (%)
Trigger Efficiency 6
Jet Identification 7
Jet Energy Scale 9
Jet Resolution 5
Taggability Scale Factor 1
b-Tag 3 14
Total
0.14
0.19
Each systematic source varied by ±1
Analysis repeated
Uncertainties dominated by Jet Energy Scale and b-tagging
Background Errors
Single Tag (%)
Double Tag (%)
Trigger Efficiency 6
Jet Identification
Jet Energy Scale 8 11
Jet Resolution 2
Taggability Scale Factor 1
b-Tag 5 12
Instrumental b-Tag 9
Instrumental Prediction 3
Cross-section
Total
0.18
0.19
Tim Scanlon (Imperial College London)

18. Sensitivity Prospects
Ingredient
Equivalent Luminosity Gain 115 GeV)
 1 fb-1 3
NN b-tagging
2.0
NN Event Selection
1.7
Di-jet Mass Resolution
1.5
Increased Acceptance
1.2
Reduced Systematics
Total 22
Largest improvement in sensitivity from
Increased luminosity
NN b-tagging
Tim Scanlon (Imperial College London)