1. Evidence for Bs Mixing and measurement of ms at CDF
S. Giagu and CDF Collaboration
University of Rome “La Sapienza”
INFN Sezione di Roma 1

2. Outline Introduction Search for Bs-Bs oscillations in CDF
Impact on the overall UT fit
Work in progress and Outlook
CDF Collaboration,
“Measurement of the Bs-Bs Oscillation Frequency”
hep-ex/ – accepted by Physical Review Letters
S.Giagu - ICHEP 2006, Moscow

3. B Meson Flavor Oscillations
Neutral B mesons can spontaneously transform in the corresponding antiparticle s,
In the SM generated via F=2 2nd order weak interactions, dominated by the exchange of a top quark
Mixing involves CKM elements,
measuring Δmq constraints
the unitarity triangle
New exotic particles may run in the loop
 mixing sensitive to NP
Form factors and B-parameters from Lattice calculations are known at ~15% level
S.Giagu - ICHEP 2006, Moscow

4. ms and the side of UT
md  f2BBB [(1-r)2+h2]  circle centered in (r,h)=(1,0)
f2BBB known at 15% from LQCD
many theoretical uncertainties cancel in the ratio
|Vts|/|Vtd| can be determined at ~4%
(hep-lat/ )
Experimental challenge:
|Vts| >> |Vtd|  ms >> md  needs to resolve > 2.3 THz oscillations
Status of Dms measurements:
LEP/SLD/CDF-I: ms > % CL HFAG Average for PDG 2006
D0 Run-II: ms  [17,21] 90% CL Phys. Rev. Lett. 97, (2006)

5. Road map to ms measurement
vertexing (same) side
“opposite” side
e,,Jet 1 4 3 2 5
Collect as many Bs as possible
Tevatron, Trigger (SVT)
Extract Signal
Bs flavor at decay inferred from decay products
Measure proper decay time of the Bs meson
L00, per event primary vertex, candidate specific decay time resolution
Determine Bs flavor at production (flavor tagging)
PID (TOF, dE/dx)
Flavor tag quantified by Dilution: D=1-2w, w = mistag probability
Measure asymmetry between unmixed and mixed events
In practice: perform likelihood fit to expected unmixed and mixed distributions

6. Event Selection: Fully Hadronic Bs
used in this analysis
Bs momentum completely reconstructed
Excellent decay time resolution, good S/N
Small BR  low statistic
Good sensitivity at high values of ms
Decay mode
Events
BsDs ()
1600
Bs Ds (K* K)
800
Bs Ds  (3)
600
Bs Ds3 ( )
400
Bs Ds3 (K*K)
200
Total
3600
Cleanest decay mode:
BsDs[] [KK] 

7. Event Selection: Semileptonic Bs
Ds Mass
Missing momentum ()
Poorer decay time resolution
Large BR  high statistic
Good sensitivity at low values of ms
l+Ds Mass
48000 l+Ds candidates, 75% are from Bs decay
Minv(l+Ds) helps reject BG
BG Sources:
Ds + fake lepton from PV
Bs,dDsDX (DslnX) cc

8. Proper decay time reconstruction
RUN EVENT D
decay B
Lxy  p PV
Decay
CDF [ps]
(stat. only)
PDG 06 [ps]
B0  D-+
1.508  0.017
1.530  0.009
B-  D0-
1.638  0.017
1.638  0.011
Bs  Ds()
1.538  0.040
1.466  0.059
Detector length scale and proper treatment of detector/selection biases controlled by performing lifetime measurements

9. Decay time resolution oscillation period @ ms=18 ps-1
Finite resolution dilutes the amplitude of mixing asymmetry:
Sensitivity maximized by making full use of all available information:
layer-00, candidate specific primary vertex and decay time resolution
Resolution measured in data in large samples of prompt D meson decays
D+ combined with prompt tracks to mimic B0-like topologies
oscillation period
@ ms=18 ps-1
M(lDs) > 3.3 GeV/c
first bin of ct
4 sampling per cycle
Hadronic decays gives CDF sensitivity at much
larger values of ms than previous experiments

10. Flavor Tagging Performances
Two types of flavor tags used in CDF
OST: produce bb pairs: find 2nd b, determine flavor, infer flavor of 1st b
calibrated on large samples of B0 ad B+ decays
SST: use charge correlation between the b flavor and the leading product of b hadronization
performances (D) evaluated in MC, after extensive comparison data VS MC
εD2 Hadronic (%)
εD2 Semileptonic (%)
Muon
0.48  0.06 (stat)
0.62  0.03 (stat)
Electron
0.09  0.03 (stat)
0.10  0.01 (stat)
JQ/SecVtx
0.30  0.04 (stat)
0.27  0.02 (stat)
JQ/Displ’d trk
0.46  0.05 (stat)
0.34  0.02 (stat)
JQ/High pT
0.14  0.03 (stat)
0.11  0.01 (stat)
Total OST
1.47  0.10 (stat)
1.44  0.04 (stat)
SSKT
3.42  0.96 (syst)
4.00  1.12 (syst)
Same-side kaon tag increases effective statistics  ~4

11. Likelihood for each event: =
Courtesy of J.Kroll
Data fitted with an unbinned likelihood function to the expected unmixed and mixed distributions
Procedure checked on B0 by fitting for md
for each event:
k=sig,bg k k k k k =
sig
pdg
ct [cm]
K-factor
isolation
Sst D
pT [GeV/c]
(*) H-G.Moser, A.Roussarie,
NIM A384 (1997)
Amplitude method(*): scan ms space: fix ms fit for A:
A consistent with 1  mixing detected at the given ms

12. Results Likelihood ratio: P-value = 0.2% (>3)
A=1 VS A=0 hypothesis
A/A (17.3 ps-1) = 3.7
hep-ex/ – accepted by PRL
P-value = 0.2% (>3)
small systematic uncertainty
dominated by knowledge of the absolute scale of the decay-time measurement
Inputs from PDG 06 and ξ= (hep-lat/ )
+0.047
-0.035

13. Impact on the overall UT Fit
SM fit
SM+NP fit
CDF
measurement
CBs = 0.97 ± 0.27
CKM fit (no Δms)
(21.5 ± 2.6) ps-1
no angles
angles only
UTfit Coll.: hep-ph/
and Vincenzo’s talk
Similar results from CKMfitter group: and Stephane T’Jampens talk

14. Work in progress BsDs+-+ (Ds +--)
Collecting new integrated luminosity
Squeezing maximum information from the data
we already have:
CDF Run II Preliminary L=1 fb-1
Systematic use of Neural Networks in signal extraction:
use decays modes previously discarded cause high BG
more signal in already used modes
Use partially reconstructed BsDs*/K and Ds:
large BR
good momentum resolution
Improve Flavor taggers:
OST: +15% D2
NN to combine OS taggers
OSKT
SSKT: ~+10% D2
better use of combined PID and kinematics
NBs = 220
BsDs+ (Ds-)

15. Summary and Outlook
CDF finds evidence for flavor oscillations in the Bs sector
Probability of a random fluctuation 0.2%
Measurement of the mixing frequency with <2% precision
Most precise measurement of |Vtd/Vts|
An important and precise experimental input for the overall test of the SM and the end of a very long effort to measure ms
… but not the end of the CDF B-physics programme

16. Random Slides
S.Giagu - ICHEP 2006, Moscow

17. Data Sample Typical inst. Luminosity 1032 cm-2 s-1
Bs candidates collected by SVT trigger
TTT: two displaced tracks
L+SVT: lepton + displaced track(s)
used in this analysis PV
d0 = impact parameter
Decay
Vertex
Typical inst. Luminosity cm-2 s-1
~1.4 fb-1 collected by CDF
~1 fb-1 (good runs) used in this analysis
S.Giagu - ICHEP 2006, Moscow

18. Other results on ms LEP, SLD, CDF-I Recent from D0 collaboration
1st direct single experiment upper bound
ms  [17,21] 90% CL
Null hypothesis probability: 5%
ms > % CL
HFAG Average for PDG 2006
D0 Coll.: Phys. Rev. Lett. 97, (2006)
S.Giagu - ICHEP 2006, Moscow

19. Example of Specific Trigger for B Physics
Level 1
- 2 XFT tracks with pT > 1.5 GeV
- opposite charge
-  < 135o
- |pT1| + |pT2| > 5.5 GeV PV
d0 = impact parameter
Decay
Vertex
Level 2
- confirm L1 requirements
- both XFT tracks
- SVT 2<15
- 120 m< |d0| <1mm
- 2o <  < 90o
- Decay length Lxy > 200m
Level 3
- confirm L2 with COT & SVX
“offline” quality track reco.
S.Giagu - ICHEP 2006, Moscow

20. Semileptonics: Correction for Missing Momentum
Reconstructed quantity
Correction Factor (MC)
Decay Time
oscillation period
@ ms=18 ps-1
S.Giagu - ICHEP 2006, Moscow

21. PID  K Separartion Power Combined PID: TOF + dE/dx
S.Giagu - ICHEP 2006, Moscow

22. Systematic Uncertainties
Hadronic
Semileptonic
related to absolute value of amplitude, relevant only when setting limits
cancel in A/A, folded in in confidence calculation for observation
systematic uncertainties are very small compared to statistical
S.Giagu - ICHEP 2006, Moscow

23. Incertezze sistematiche su ms
systematic uncertainties from fit model evaluated on toy Monte Carlo
have negligible impact
only relevant systematic: knowledge of lifetime scale
Syst. Unc
All other syst.
< 0.01ps-1
SVX Alignment
0.04 ps-1
Track Fit Bias
0.05 ps-1
PV bias from tagging
0.02 ps-1
Total
0.07 ps-1
All relevant systematic uncertainties are common
between hadronic and semileptonic samples
S.Giagu - ICHEP 2006, Moscow

24. Amplitude Scan: Hadronic decays
data period 1
data period 2
data periodo 3
S.Giagu - ICHEP 2006, Moscow

25. Amplitude Scan: Semileptonic decays
data period 1
data period 2
data period 3
S.Giagu - ICHEP 2006, Moscow

26. Parameterization of the tagging decision
Exploit peculiarity of each tagger to minimize mistag probability
example: soft muon tag
 from b decay
jet axis 
ptrel
 from c decay
S.Giagu - ICHEP 2006, Moscow

27. SSKT Calibration
Dilution measured in high statistic samples of light B meson decays and compared with the results of simulation
Dominant source of systematic uncertainty: Data/MC agreement ~O(14%)
S.Giagu - ICHEP 2006, Moscow

28. Negative log likelihood ratio
S.Giagu - ICHEP 2006, Moscow