1. Studies of the decay B  c K at the BaBar experiment Nick Barlow University of Manchester December 2003

2. 2 The BaBar Collaboration The BaBar collaboration consists of –Nearly 600 physicists –From 10 countries –75 Institutions The  c analyses presented here were mainly performed by: –Me –Witold Kozanecki (Saclay) –Stefania Ricciardi (RHUL) –Frank Jackson (Manchester) –Gautier Hamel de Monchenault (Saclay)

3. 3 Outline The PEP-II B-factory, and BaBar detector CP violation in the quark sector of the S.M. –CKM matrix and Unitarity Triangle –3 Types of CP violation in B decays Measuring the CKM parameter sin2  using neutral B decays to charmonium + K. (Specifically B  c K S ) –Reconstructing B candidates –B flavour tagging –Time-independent and time-dependent fits

4. 4 Motivation for the BaBar experiment CP violation is interesting because –It is one of the least tested areas of the Standard Model –Has some relevance to the matter/antimatter asymmetry in the Universe Before BaBar and Belle, CP violation had only been observed in the kaon system –Hadronic uncertainties make comparison with Standard Model parameters difficult High luminosity of PEP-II facilitates numerous other precision measurements in B physics (and tau physics…)

5. 5 The PEP-II B-factory e + e -  (4S)  BBbar 9 GeV electron beam 3.1 GeV positron beam Boost  (4S) in lab frame:  0.55

6. 6 PEP-II performance Design peak luminosity = 3.0*10 33 cm -2 s -1 Record peak luminosity = 6.8* 10 33 cm -2 s -1 24 hr record integrated = 484.3 pb -1 7 day record integrated = 2.49 fb -1 Records improving all the time, largely thanks to trickle injection… Results in this talk use 81fb -1 (Runs 1 and 2).

7. 7 Trickle injection Inject charge continuously into LER, while BaBar is taking data Will increase integrated luminosity by at least 15% –If backgrounds are under control…

8. 8 The BaBar detector e + (3.1GeV) e - (9GeV) Silicon Vertex Tracker Drift CHamber Electro- Magnetic Calorimeter Detector of Internally Reflected Cherenkov light Instrumented Flux Return 1.5T solenoid

9. 9 Tracking and Vertex finding SVT consists of 5 layers of double-sided silicon strip detectors –Single hit resolution ~80 microns. DCH consists of 40 layers (10 superlayers) of wires, arranged in axial and stereo layers in helium/isobutane gas mixture –Measures momentum and position of charged particles –  (p T )/P T =0.13%P T  0.45%

10. 10 Particle Identification Charged hadrons ( , K) identified using dE/dx in the DCH and SVT, and Cherenkov angle in the DIRC –K-  separation>3.4  for P<3.5GeV/c Leptons identified using the IFR (  ) and EMC (electrons)

11. 11 Neutral particles EMC consists of 6480 CsI(Tl) crystals, arranged in a barrel+endcap configuration. Measures energy of electrons and photons over a wide energy range.

12. 12 CKM matrix and Unitarity Triangle Quark mixing is described by the CKM matrix Wolfenstein parametrization Complex phase  CP violation Unitarity condition: Can be represented as a triangle on complex plane: Im Re1 (,)(,)   

13. 13 Unitarity triangle cont. Angles are given by: Im Re1 (,)(,)   

14. 14 Existing constraints on Unitarity Triangle Errors are dominated by theoretical uncertainties! Constraints come from - CP violation in K 0 mesons - V ub and V cb measurement - B d and B s mixing http://ckmfitter.in2p3.fr

15. 15 3 Types of CP violation in B decays 1.CP violation in decay (direct CP violation) –Amplitude for a decay is not same as amplitude for CP conjugate decay 2.CP violation in mixing (indirect CP violation) –Mass eigenstates cannot be chosen to be CP eigenstates 3.CP violation in interference between decays with and without mixing –Time-dependent asymmetry in decays of and to common CP eigenstate. CP mixing decay

16. 16 CP violation in neutral B decays Consider neutral B decays into a CP eigenstate f CP that is accessible to both B 0 and B 0 : Can define measurable quantity :

17. 17 CP violation in neutral B decays Time dependent asymmetry given by: If | | = 1 (no direct CP violation), this reduces to:

18. 18 B  Charmonium K 0 For B  Charmonium K 0, both ‘tree’ and dominant ‘penguin’ diagrams have same weak phase, so no direct CP violation, | | = 1 d c d BB J/  c KK b s c d c d b c s =  chK sin2 

19. 19 B  Charmonium K 0 B  J/  K S is the ‘Golden channel’ at BaBar The J/  decays to leptons – clean reconstruction with good efficiency The  c has a much larger intrinsic width than the J/ , and only decays hadronically Need to work harder to reduce background Less than half the total width is accounted for by known decay channels Although J/  and  c have opposite intrinsic CP, the J/  is a vector, so (-1) L term in CP of 2-body state means that B  J/  K S and B  c K S decays have the same CP eigenvalue (-1)

20. 20 Experimental method BBbar pair evolve as coherent P-wave state. –Always exactly 1 B 0 and 1 B 0 Boosted wrt lab frame: can measure distance  z between B decay vertices. e+e+  (4S) Fully reconstructed B decay (CP eigenstate) e.g. B  J/  K s, B   c K s Tag flavour of B- meson using e.g. charge of a primary lepton  t   z/(  c) B tag B CP e-e-

21. 21 Time-dependent asymmetry In principle, just measure time difference  t for events where B tag is tagged as a B 0 or a B 0. In practice, tagging and  t determination are not perfect – need to modify Probability Density Functions (PDFs) with mistag fractions (w) and resolution functions. B 0 tags

22. 22 Time dependent PDF P(  t)  exp(–|  t|/  B ) ( 1 ± (1-2w)  CP sin2  sin(  m  t) )  R(  t) dilution Resolution function Can extract sin2b from time-dependent fit to  t distributions of tagged events, BUT Need to know mistag fraction and resolution function Need to add other PDFs for background events, which in general will have different time dependence CP eigenvalue

23. 23 Neutral B-meson mixing Neutral B mesons can mix –B tag and B flav can be opposite flavour (no mixing) or same flavour (mixing) e+e+  (4S) Fully reconstructed B decay (flavour eigenstate) e.g. B  D -  + Tag flavour of B- meson using e.g. charge of a primary lepton  t   z/(  c) B tag B flav e-e-

24. 24 B-mixing (2) PDF for B-mixing is given by: Resolution function Dilution = 1-2w, where w is fraction of wrongly tagged events Real probability of B-mixing unmixed mixed

25. 25 B flavour tagging Fully reconstructing B tag not practical – would be too inefficient. Instead, use Neural Net tagging algorithms based on different physics processes –Primary lepton from W in b  c transition –Charged kaons from b  c  s transition –Charge of slow pions from D* decays –Charge of highest p T track Divide into 4 hierarchical, mutually exclusive ‘Tagging Categories’, –Lepton, Kaon1, Kaon2, Other Each tagging category has its own mistag fraction,  t resolution function, measured on real data using B mixing sample.

26. 26 Tagging performance Important parameter is effective tagging efficiency Q=eff(1-2w) 2 –This is what goes into the error on sin2b CategoryEfficiency  (%) Mistag  (%) Quality Q(%) Lepton 10.2  0.22.7  0.3 9.2  0.2 Kaon1 17.8  0.28.8  0.412.1  0.3 Kaon2 16.3  0.220.4  0.6 5.7  0.2 Other 22.7  0.231.3  0.6 3.2  0.2 Total 67.1  0.530.2  0.5

27. 27 Measuring  t Resolution on  t is dominated by determination of z decay position of B tag –Use constraints from knowledge of the beamspot position, and the momentum of B rec  t=  z /(  c) + small correction due to small momentum of Bs in Y(4S) frame

28. 28  t resolution Different PDF used for each tagging category –All use triple Gaussian (core, tail, outliers), and event-by-event errors. –  t resolution usually better for lepton category 8 free parameters: Relative fractions of tail and outliers Scale factor for width of core Gaussian Bias for tail Gaussian Bias factors for core Gaussians in each tagging category

29. 29 Data samples B CP sample – reconstructed B 0  c K S events –Used for measuring sin2b B flav sample – events where 1 B is reconstructed into a flavour eigenstate e.g. D *0  - –Used for measuring mistag fractions and  t resolution B sig+ sample - reconstructed B +  c K + events –Used for cross-checks, higher statistics than neutral mode. Numerous Monte Carlo samples –Used for cross-checks, and tuning the fit – can vary generated value of sin2b.

30. 30 Reconstruction of B  c K candidates Use decay channels  c  K S K +  - and  c  K + K -  0 Most BaBar (and Belle, CLEO) B analyses use the variables m ES and  E, which are (nearly) uncorrelated. –  E is the measured – expected energy of the reconstructed B candidate –m ES is the beam-energy substituted B-mass (assumes  E=0) m ES (GeV 5.35.2

31. 31 Fighting background First run a ‘skim’ – loose cuts on topological variables, charged track multiplicity etc. to reduce dataset to a manageable size Apply cuts on reconstructed masses of K S,  0 etc. –All cuts optimized to maximise S/sqrt(S+B) Estimated from signal MC and  E sidebands of on-resonance data Use a ‘Fisher Discriminant’ to fight background from u,d,s,c quark events –Linear combination of ‘event shape’ variables –Light quark events tend to have jet-like structure in CM frame, while BBbar events tend to be more spherical, as B mesons are produced almost at rest in CM frame

32. 32 Types of background Combinatorial background, from light quark events or BBbar events where B rec is reconstructed using tracks from both Bs –Can be studied using off-resonance data, MC, and  E sidebands of on-resonance data –Modeled using a threshold function (ARGUS function) in m ES Peaking background from BBbar events –In our case, studies on high-statistics BBbar MC indicate that this is dominated by decays to the same final state as our signal, but without an intermediate  c resonance: Neither of these backgrounds is expected to have any structure in m X – the mass of the reconstructed  c candidate. B+B+ K+K-0K+K+K-0K+ CC B+B+ K *+ 

33. 33 Time independent fit Extract signal and background yields using 2D unbinned maximum likelihood fit to m ES and m X  c signal: Breit- Wigner  Gaussian2 in m X Gaussian1 in m ES J/  signal: Gaussian2 in m X Gaussian1 in m ES Peaking background: Gaussian1 in m ES linear in m X Combinatorial background: Argus shaped in m ES, linear in m X

34. 34 Results of time independent fit KsKPi neutral channel: Tagging category Fraction of CP signal Fraction of peaking bg Lepton0.38 ±0.090.18±0.06 Kaon 10.11±0.030.04±0.01 Kaon 20.16±0.040.05±0.02 Other0.12±0.030.03±0.01

35. 35 The sin2  fit The value of sin2  is obtained from a simultaneous fit to the B flav and B CP samples –Takes into account potential small correlations between sin2b and resolution functions etc., and uses statistical power of B CP sample. Altogether there are 34 free parameters –Mainly due to resolution functions for each tagging category for signal and background, and the composition and time-dependence of the backgrounds in both the B CP and B flav samples.

36. 36 Results on MC B 0  c K S signal, generated with sin2b=-0.7 sin2b=-0.744±0.074 B +  c K + signal, no CP violation expected sin2b=-0.062±0.059

37. 37 Results on B sig+ sample No CP violation expected: sin2b=0.11±0.16

38. 38 Results on B CP sample i.e. the actual result! sin2b=0.51±0.33

39. 39 More cross-checks Can use ‘Toy MC’ experiments, where events are generated to according to a PDF, to check validity of result: –Compare maximised likelihood found in fit to data with distribution of values found in 500 Toy MC experiments –Compare statistical error in fit to data with those found in 500 Toy MC experiments

40. 40 Systematic errors Source Error on sin2  Fraction of peaking bg0.02 CP of peaking bg0.04 Dilution &  t resolution 0.03 Combinatorial bg description0.03  (m ES ) 0.0015  (m X ) 0.02  m B  B (PDG 2000) 0.01 Total systematic error0.08

41. 41 Comparison with other decay modes  CP =-1 CP =+1 sin2  = 0.755  0.074sin2  = 0.723  0.158

42. 42 Constraints on the Unitarity triangle http://ckmfitter.in2p3.fr B  c K S only All BaBar results

43. 43 Conclusions BaBar is a good place to do B physics –Very high luminosity of PEP-II –Clean environment of e + e - collisions BaBar and Belle have made precision measurements of CP violation in B mesons –Values of sin2b agree well with S.M. expectations It is (just about) possible to measure sin2b using the decay channel B  c K S –Helped reduce the statistical error by 0.01 for the last publication!! –Studies on MC indicate it will be advantageous to include this channel in the main sin2b measurement for the forseeable future

44. 44

45. 45 Backup slides

46. 46 What is an  c ?  c is the ground state of the charmonium system –Mass = (2979.7  1.5) MeV/c 2 (PDG 2002) –Width is very poorly known: –BaBar has just made a new measurement using  c from 2 photon production: Use J/psi from ISR to measure detector resolution This is the value we use in our analyses