110 likes | 219 Vues
Search for Top Flavor Changing Neutral Current Decay t → qZ. Ingyin Zaw DOE Review August 21, 2006. Why?. BR(t →qZ) ≈ O(10 -14 ) in the Standard Model No tree level, only higher order diagrams, e.g. penguin diagrams New Physics models predict branching ratios up to O(10 -2 )
E N D
Search for Top Flavor Changing Neutral Current Decay t→qZ Ingyin Zaw DOE Review August 21, 2006
Why? • BR(t→qZ) ≈ O(10-14) in the Standard Model • No tree level, only higher order diagrams, e.g. penguin diagrams • New Physics models predict branching ratios up to O(10-2) • SUSY, Higgs doublet, Warped extra dimensions, etc. • Any observation at the Tevatron →New Physics • Have large sample of top pairs • ~1 fb-1 of data • ~7k top pairs produced Penguin Diagram New Physics?
Signature • One top decays via FCNC (t→qZ), the other via SM (t→Wb) • Z ℓ+ℓ- / νν / qq, Wℓν / qq (ℓ=e/μ)
Signal Events with a Z Pythia MC • ttWb qZ ℓ+ℓ- + ≥4 jets • Reconstruct Z from e+e- or μ+μ- • 76 GeV/c2<MZ<106 GeV/c2 • Require ≥ 4 jets • Jet Et ≥15GeV, corrected • Jet |η| < 2.4 • One jet is a b jet → can have a secondary vertex tag • Blind analysis • Signal region: Z+≥4jets • Use both b tagged sample and anti-b tagged sample • BRAcceptanceEfficiency (before tagging): 0.95% • Event tagging efficiency (≥ 1 tag): 52% 28% 72% Control Region Signal Region Number of Jets jet tagged b jet l l jet jet
Backgrounds • Z+Jets • Standard Model Top • Others (small) • Dibosons (WW, WZ, ZZ) • From MC • 3.5 events before tagging, 0.5 tagged • Fake Zs • Estimate from same-sign Zs
Z+Jets, Pretag • Use both data (700 pb-1) and Monte Carlo to estimate background • Leading order Alpgen+Pythia Monte Carlo • Correctly models the Z+njet distribution • Absolute Z cross-section is low • Normalize to inclusive Zs in data • Expected background before tagging for 1 fb-1 of data: • Zs in data (scaled from 700 pb-1): 79500 • Z+≥4j fraction in MC: 0.156% • ~ 125 events in signal region Number of Jets
Z+Jets, Tagged • Z+Heavy Flavor • Measure heavy flavor fractions and tagging efficiencies in MC • Z+bb: • Fraction: ~9.5% • Tagging efficiency: ~48% • Z+cc: • Fraction: ~17% • Tagging efficiency: ~14% • Predicted tag rates checked in control region in data • HF fractions may be higher in data than MC • See Joao’s talk on W+HF and K-factor • Z+Light Flavor (Mistags) • Apply mistag templates from data to MC • Mistag rate: ~6% • Expected background: ~6 events • Expected tagged background: ~18 events
Dilepton Mass (GeV) Standard Model Top SM top MC (170 fb-1) • Estimated from Monte Carlo simulation with Pythia • tt→WbWb→lν lν bb events with dilepton mass in Z mass window • tt→WbWb→lν jj bb events where a jet fakes a lepton • Expected background in 1fb-1: • ~ 4 events before tagging • ~ 3 events after tagging With ≥ 4jets Events with a Z Candidate 91% 9%
Feldman-Cousins Limits and Optimization • Limit Calculation: • Assume no signal • Expected upper limit for signal: • i = number of observed events • nb = number of expected background events • Optimization: • Optimize cuts for best limit • Variables: • jet ETs • HT = • Masses • FCNC top mass • W mass • SM top mass Expected Limit vs. HT HT (GeV)
Sensitivity • Still optimizing for best limit • No systematic uncertainties yet • Expected limit at 95% C.L. (assuming no signal): • Anti-tagged sample: 23% • Dominant background: Z+Jets • Tagged sample: 18% • Dominant background: Z+Jets, SM top • Combined: 10% • Previous limits: • 33% (CDF Run I), direct search • Same channel as our search • No b-tagging • 13.7% (LEP), indirect search • Single top production via flavor changing neutral current
Plans • Optimize event selection • Finish cross-checks with data • Incorporate systematic uncertainties • Results! (hopefully, very soon)