B s Mixing Results for Semileptonic Decays at CDF Vivek Tiwari Carnegie Mellon University on behalf of the CDF Collaboration

2 B Meson Flavor Oscillations Neutral B mesons can oscillate into their corresponding antiparticles via 2nd order weak interactions, dominated by the exchange of a top quark Several theoretical uncertainties cancel in the ratio New Physics may affect  m s /  m d  New particles in the loop V ts  = (hep/lat )

3 Neutral B Oscillations Neutral B Meson system Mixture of two mass eigenstates: B H and B L may have different mass and decay width   m = m H – m L   =  H -  L In case of  = 0

4 B Physics at the Tevatron All B hadrons produced:  B +, B d, B s, B c,  b … Large B cross section  Tevatron:  B Factories: However, the total inelastic cross section,  (total  is more than 1000 times bigger  Need to select B events with high purity  It’s all about triggers at hadron colliders

5 Tevatron Performance Delivered luminosity ~ 2.0 fb -1 (~1.6 fb -1 on tape) Mixing measurements at CDF use ~ 1.0 fb -1 Tevatron regularly making new records  Peak initial luminosity ~2.3 x sec -1 cm -2  Record weekly integrated luminosity ~ 33 pb -1 Delivered : 1983 pb -1 Collected : 1606 pb -1 Used in this analysis

6 The CDF II Detector Excellent momentum resolution  (p)/p<0.1% Large B yields:  High rate trigger/DAQ Particle Identification:  TOF, dE/dX in COT  Calorimeter & muon chambers Proper time resolution  Silicon detectors: SVXII, L00

7 B Physics Triggers at CDF Conventional di-muon (J/  ) trigger  p T (  ) > 1.5 GeV  Samples used for flavor tagging studies Lepton + displaced track (SVT)  Lepton = e,  with p T > 4.0 GeV  p T > 2 GeV displaced track (120  m < I.P. (track) < 1mm)  Large semileptonic samples for mixing and flavor tagging studies Two displaced tracks  Two p T > 2 GeV SVT tracks  Provides access to hadronic decays and large semileptonic samples with lower p T leptons

8 Overview of the Measurement “same” side “opposite” side Reconstruct B s decays (determine decay flavor from decay products) Measure proper decay time of the B s mesons Infer B s flavor at production (flavor tagging) e,  e+e+ LTLT LTLT

9  m s Measurement Significance B s mesons mix much faster than B d The measured asymmetry is diluted by mistags, since the initial state flavor is not perfectly known  Oscillation Amplitude: D=1-2w, w = mistag probability Signal/Background Proper time resolution Effective tagging power (  =tagging efficiency) Moser, Roussarie, NIM A384 (1997)

10 Particle Identification at CDF Lepton Identification  Combine variables into a global likelihood to discriminate against fake leptons  Electron: Calorimeter, shower & pre-shower quantities and dE/dx  Muon: Track-muon matching and calorimeter variables Likelihood based id is used for semileptonic signal selection as well as opposite side flavor tagging

11 Particle Identification at CDF (contd.) Charged Kaon Identification  Combine information from dE/dx and TOF  dE/dx provides ~ 1.5  separation for p > 2.0 GeV tracks with 100% efficiency  TOF provides ~ 2.0  separation for p < 1.5 GeV tracks with 60% efficiency Used for B s signal selection and in the same side & opposite side kaon tagging algorithms

12 B s Signal Reconstruction in Semileptonic Decays Semileptonic B s decays  B s  l D s X reconstructed in three final D s states: D s   / K*K /   l = e,  collected via the two- track and l +SVT triggers  Characterized by large branching ratios  Incomplete reconstruction (missing neutrino and other neutral particles)

13 B s Signal Reconstruction (contd.) In D s   (  K  K  ) & D s  K*K (K*  K    ) modes, kaon identification is used  Helps suppress combinatorial background composed largely of pions  Helps reduce reflection from D   K       in D s  K*K mode Physics backgrounds contamination ~ 20-25%  Depends on lepton momentum  Split sample into cases when lepton is a trigger track Total B s  l D s X signal yield is 61,500 l D s : D s   29.6 K l D s : D s  K*K 22.0 K l D s : D s   9.9 K

14 B s Signal Reconstruction (contd.) Mass ( l -D s ) distribution  Helps discriminate against physics, fake lepton & combinatorial background Obtain estimate of fake lepton background ~ 5-10% Mass ( l -D s ) distribution for fake leptons obtained via anti-selection on lepton likelihood  Quantifies missing momentum for signal B s  l D s X candidates Crucial for maintaining sensitivity at higher values of  m s

15 Proper Decay Time Reconstruction Trigger distorts decay time distribution  Correct using efficiency function obtained from trigger simulation on Monte Carlo Missing decay products  Correct statistically using a missing momentum factor (k-factor) where distribution of is obtained from Monte Carlo “ Trigger” turnon pattern limit |d 0 | < 1 mm

16 Proper Decay Time Resolution Excellent decay time resolution critical for sensitivity at high  m s Sensitivity in semileptonic decays is driven by low decay time or high Mass( l -D s ) candidates  Variation of k-factor with Mass( l -D s ) significantly improves decay time resolution  Exploited by using Mass( l -D s ) directly in the fit.  ct determined directly from data  Event-by-event  ct is used taking into account dependence on kinematical variables like isolation, opening angle as well as vertex   (more details in Jeff Miles’ talk) Osc Freq 18 ps -1

17 B Flavor Tagging (Opposite Side) b quarks produced in pairs: use the other B to infer production flavor  Lepton (e/  Tagging: Semileptonic decay of OS B (high purity/low efficiency)  Kaon Tagging: Kaon from OS b  c  s transition (medium purity/medium efficiency)  Jet Charge Tagging: Weighted sum of fragmentation and decay products of OS B (low purity/high efficiency) Issues  OS B not always in acceptance  OS B mixing diminishes tagging performance

18 B Flavor Tagging (Opposite Side contd.) Combine tagging algorithms using a Neural Net  Use dependence of expected tag purity on particle-id / kinematical variables Apply the combined tagging algorithm on samples of B + and B d decays  Calibrate expected dilution  Cross-check of the complicated unbinned maximum likelihood fit framework Combined tag  Measured value of  m d consistent with PDG

19 Charge of closest fragmentation track correlated to B production flavor  Superior to OS tagging due to better acceptance and doesn’t suffer from OS mixing SSKT performance cannot be determined from B s data  Rely on Pythia MC prediction Tagging track identification based on a NN combination of kaon probability and kinematical variables SSKT B Flavor Tagging (Same Side Kaon Tagging)

20 Fourier Analysis Technique Two domains to fit for oscillations:  Time: fit for cosine wave  Frequency: examine spectrum Time Domain Approach  Fit for  m s in p(t)~(1 ± D cos  m s t)  Good for measuring  m s Frequency Domain Approach  Fit for A(  m s ) in p(t)~(1 ± A D cos  m s t)  A = 1 for true  m s, else A=0  Good for exclusion, combining measurements Moser, Roussarie, NIM A384 (1997)

21 Semileptonic Amplitude Scan Combined sensitivity on 1 fb -1 : 19.3 ps -1 Amplitude is consistent with unity ~17.8 ps -1 (A/    Points: A±  (A) from likelihood fit for different  m s Green band: A±1.645  (A) Dashed line:  (A) as function of  m s Measurement sensitivity:  (A) = 1

22 Likelihood Profile Evidence of oscillations  Likelihood global minima at  m s = 17.9 ps -1 Strict Gaussian interpretation of the minima is not possible but ±1  around the minima gives an error on  m s ~ 0.3 ps -1 Can also set a 95% double bound:  m s  [16.9,19.5]

23 Conclusions World’s best sensitivity in B s semileptonic decays =19.3 ps -1 Evidence of oscillations at  m s = 17.9 ps -1 95% double bound:  m s  [16.9,19.5] Details on mixing in hadronic decays at CDF and combination with semileptonic decays: see Jeff Miles’ talk

24 Slides for Reference

25 Systematic Uncertainties