B0%20mixing%20at%20CDF

1
B0 mixing at CDF

• Marco Rescigno INFN/Roma

• Bs Mixing predictions
• Tevatron CDF
• CDF analysis
• Collecting B(s)
• Flavour Tagging
• Measuring Dmd
• Dms limit
• Tevatron Sensitivity extrapolation

2
B0 Flavour Oscillations
Flavour oscillations occur through 2nd order weak
interactions
Dmd (exp.) 0.5100.005 ps-1 (HFAG
2005) Lattice-QCD f2BdBBd (2233312)
MeV f2BsBBs (27638) MeV ? Vtd
determined at 15 But in the ratio uncertainties
cancels
Measuring Dms/Dmd tests Vts/Vtd with 5
theory error
x 1.240.040.06
3
Unitarity Triangle Dms

• Brown Band Dmd measurement 15 uncertainty
• Dashed circle lower limit on Dms /Dmd ? Upper
  Limit on Vtd
• The lower bound on Dms already gives a
  constraint to Unitarity Triangle
• http//utfit.roma1.infn.it

CKMfitters version
from Dmd/Dms
4
CKM fit
SM CKM-fit prediction for Dms
Input from Vub/Vcb, ? md, eK, sin2ß, cos2ß, a
and ?
?ms 20.4 2.8 ps-1 15.1, 26.3 _at_ 95 CL
include also ?ms limit
Dms 18.9 1.7 ps-1 15.7, 23.0 _at_ 95 CL
A very narrow shooting range for collider
experiments!
New Physics _at_ 3s for Dms gt 31 ps-1
b-s sector much less constrained (yet) than
b-d Large New Physics contribution to Bs mixing
and its phase still possible!
compatibility plot
5
Outline/RoadMap to Dms
Recall expression for significance in a mixing
measurement
5) Maximize tagging rate x dilution2 Know your
mistag rate from a Dmd measure Need statistic!
1) Trigger design to maximise Signal (S)? highest
BandWidth Fight for your Band Width (if in a
general-purpose exp.) Keep your trigger alive!
2) Good momentum (mass) and energy (!) resol. for
max S/(SB)
Amplitude scan
3) Measure tB on semileptonic hadronic on
multiple triggers (with/without lifetime bias)

• 4) Improve st
• fully reconstructed !
• L00 close to beam pipe
• Primary vertex resolution.

6
Collecting data
7
Luminosity Delivered/Recorded
1fb-1 delivered by Tevatron Run II May 23rd
2005 !!
500 pb-1 on tape Fall 04 shutdown
Present result based on 360 pb-1 (100 pb-1
lost due to drift chamber ageing problem, now
solved)
8
Silicon Vertex Tracker

• Triggering on displaced vertex at CDF using SVT
  main novelty in Run II, workhorse for CDF
  B-physics program. See at this conference
• Charmless decays (Donati)
• SVT trigger (DellOrso)
• CDF way to get fully reconstructed decays useful
  for mixing (and other good stuff)

• Main Trigger requires
• 2 opposite charge tracks,
• Pt ? 2 GeV/c,
• impact parameter d0 gt 120 µm
• Scalar pt sum gt 5.5 GeV/c
• Projected decay length Lxy gt 200 µm
• 2 lt ?? lt 90
• Add a dynamically prescaled LOWPT trigger with
• no opposite charge and no Pt sum to fill
  available bandwidth at low luminosity

9
Two different Bs signatures
Fully reconstructed HADRONIC modes

• Complete momentum reconstruction
• Good proper time resolution
• High Bs mass resolution ? high S/B
• Selected by Two Track Trigger (SVT)
• Two displaced tracks (w large SVT Impact
  parameter)
• LOW statistics (useful BR)
• Demanding on L1 B/W gt30 KHz _at_ 1E32cm-2s-1
• 10 KHz _at_
  5E31cm-2s-1

p
Ds
Bs
P.V.
Partially reconstructed SEMILEPTONIC modes

• Missing momentum carried by the n
• Visible proper time corrected from MC (K factor)
• Proper time resolution diluted by missing
  momentum
• Cannot reconstruct Bs mass ? different S/B
• Selected by dedicated trigger (lSVT)
• One displaced tracks (w large SVT Impact
  parameter)
• One Lepton m,e (pT gt4 GeV/c)
• HIGH statistics and well behaved trigger

m,e
n
Ds
Bs
P.V.
10
Hadronic Bs signals
NBS 52633 S/B 2 sM ? 15 MeV
Satellites
( Not used in this analysis )
Other signals
11
Semileptonic Bs Signals
Other signals

• Missing PT No Bs mass peak
• Use Ds mass signals
• Charge correlation between l and Ds
• l Ds- Right-sign signal
• l - Ds- Wrong-sign background
• Right-sign peak is not pure signal
• 20 background
• Ds fake lepton from primary
• B0,B g Ds D X with D g l nX
• c-c backgrounds

12
Signal Yields Summary
(S/B)
BsgDsp Dsgfp 52633 (1.8)
BsgDsp DsgKK 25421 (1.7)
BsgDsp Dsgppp 11618 (1.0)
BgD0p D0gKp 6200
B0gDp- DgD0p 2800
B0gDp- DgKpp 5600
Hadronic Bs modes 900 events
O(104) B0/B calibration modes
(S/B)
Bsgl Ds Ds g fp 435594 (3.1)
Bsgl Ds Ds g KK 175083 (0.4)
Bsgl Ds Ds g ppp 157388 (0.3)
Bgl D0 D0g Kp 100K
B0gl D DgD0p 25K
B0gl D Dg Kpp 52K
Semileptonic Bs modes 7700 events
O(105) B0/B calibration modes
13
Measuring ct
14
Decay Time Bias
Extract proper time at decay from B flight
distance in the transverse plane

• Two complications
• Trigger bias on Lxy
• Correct for missing ? in semileptonic

• Trigger and reconstruction requirements affect
  Lxy
• Trigger (impact parameter) cuts at low ct
• SVT acceptance at high ct
• ct efficiency from Monte-Carlo
• B production/decay model
• detailed Trigger/Detector simulation
• Test with high-statistic B0/B samples

15
B0 and B hadronic modes ct
B
B0
16
Hadronic Modes Lifetime
ct cm
mB GeV/c2
(stat) (syst)
Systematic summary
t(B) 1.6610.0270.013 ps t(B0)
1.5110.0230.013 ps t(Bs) 1.5980.0970.017ps
HFAG 04 average
SVT bias syst. small
t(B) 1.653 0.014 ps t(B0) 1.534 0.013
ps t(Bs) 1.469 0.059 ps
17
Semileptonic Bs Modes Lifetime
Introduce K factor p(l Ds)/p(B) to account for
missing ?
Real Ds backgrounds prompt and physics
t 1.5210.040 ps
Combined l-Ds lifetime result 1.4770.032 ps
stat. err .only (analysis ongoing) HFAG 05
flavour specific 1.472 ? 0.045 ps (DØ 05
Dsl 1.4200.0430.057 ps)
18
Effect of proper time resolution

• The amplitude of mixing asymmetry is diluted by
  a factor

Vertex resolution (constant)
Momentum resolution (proportional to ct)
Dm 15 ps-1 st167 fs sp /p15
Dm 15 ps-1 st100 fs Dst 0.32
Semileptonic mode like
Hadronic mode like
19
Bs decay time resolution
Huge prompt (90) Ds track sample to correct
sct error calculation and parameterize as a
function of several variables.

• Hadronic
• ltsct0gt 30 mm (100 fs)
• sp/p lt 1
• Semileptonic
• ltsct0gt 50 mm (167 fs)
• sp/p 15 (K factor
• due to missing neutrino)

20
Flavour Tagging
21
Flavor Tagging

• Same side tagging
• Use fragmentation track
• B0, B, and Bs are different
• Kaon around Bs PID is important (more at the
  end of the talk)
• Opposite side tagging (5 algo)
• Use the other B in the event
• Semileptonic decay (b g l-)
• (1) Muon, (2) Electron
• Use jet charge (Qb -1/3)
• (3) Jet has 2ndary vertex
• (4) Jet contains displaced track
• (5) Highest momentum Jet

Kaon
Used only Opposite Side Tags so far for Bs
22
Calibration Sample for Taggers

• Use inclusive semileptonic decays from the
  leptontrack trigger (gt106 events)
• Lepton charge gives true B flavour
• Tag the other b

• Need high stat. sample to develop and calibrate
  tagging algorithm
• High purity reached after leptontrack mass cut
  applied
• Statistical Power of a tag eD2
• Tagging efficiency (e)
• Tagging dilution (D 1-2w)
• w mistag rate
• Parameterize dilution as a function of relevant
  variables and wheight events with their
  event-by-event dilution
• Dividing events into different classes based on
  tagging power improves combined eD2
• Calibration of the tagger performance requires
  high statistics!

23
Flavor tagging Soft Leptons
Run I
Run II
Likelihood based electron and muon ID Using
combination of calorimeter,muon detector,dE/dx
info Similar performance as in Run I
(eD20.90.1 ) Dmax0.4 ? 30 mistag rate
Tag type eD2 ()
Muon (0.700.04)
Electron (0.370.03)
2ndary vtx (0.360.02)
Displaced track (0.360.03)
Highest p jet (0.150.01)
Total (exclusive) 1.6
24
Flavor tagging Jet Charge
Run II
Run I

• Cone based jet algorithm compute Jet Charge of
• Secondary Vertex tagged jets
• Jet Probability tagged jet
• Highest P jet
• Similar performance as in Run I (eD20.80.1 )
• Dmax0.4 ? 30 mistag rate

Tag type eD2 ()
Muon (0.700.04)
Electron (0.370.03)
2ndary vtx (0.360.02)
Displaced track (0.360.03)
Highest p jet (0.150.01)
Total (exclusive) 1.6
25
B0 mixing and dilution scaling

• Validation of the flavor tag calibration using
  B0 and B sample
• B0 g Dp, B g D0p
• B0 g J/yK0, B g J/yK
• Fit the Dilution scale factor S
• 1 if the tag calibration is correct.
• 5 scale factors for 5 tag types
• Effective Dilution depend on detail of the
  samples (e.g. Pt spectra)
• Scale factors are then used for Bs mixing
  analysis for hadronic channels
• Same thing for semileptonic decays

B0 all Tags
B all Tags
26
B0 mixing results
HADRONIC SEMILEPTONIC
Dmd (0.5030.0630.015) ps-1 (0.4980.0280.015) ps-1
Total eD2 (1.120.23) (1.430.09)
Muon 0.830.100.03 0.930.040.03
Electron 0.790.140.04 0.980.060.03
Vertex 0.780.190.05 0.970.060.04
Track 0.760.210.03 0.900.080.05
Jets 1.350.260.02 1.080.090.09
Dilution scale factor

• Dmd consistent with WA 0.5100.005 ps-1
• Total eD2 1.11.4
• All dilution scale factors consistent with 1
• Hadronic 1525 uncertainty
• Semileptonic 515 uncertainty

27
Dms scan
28
Amplitude Scan for B0d(s)

• Introduce Amplitude in Likelihood
• Amplitude scan
• Fit the amplitude for fixed Dm
• Amplitude A, uncertainty sA
• Repeat the fit for different Dm
• Amplitude will be consistent with
• 1 if mixing detected at the frequency Dm
• 0 if there is no mixing
• Example for B0 Hadronic sample
• Amplitude 1 at Dm 0.5 ps-1
• Amplitude 0 at Dm gtgt 0.5 ps-1

• HFAG 04
• 95 CL limit is Dms gt 14.5 ps-1
• Sensitivity 18.2 ps-1

Hadronic B0 sample 8K ev.
Mix at Dmd
29
Amplitude Scan result
semileptonic
Hadronic has no sensitivity (yet) but is better
behaved at high Dms
Systematic errors are negligible with respect
to statistical in both cases
details
30
CDF/World Comparison
CDF2 B (hadronic) 9th best _at_ Dms10ps-1 ? 5th
best _at_ Dms19ps-1 180 ? 60 worse sensitivity
than best experiment CDF2 Dl (semil.) 7th best _at_
Dms10ps-1 ? 8th best _at_ Dms19ps-1 130 ? 95
worse sensitivity than best experiment
Stat. only!
31
CDFWorld Combined Result

• World Average CDF Run II
• Sensitivity 18.6 ps-1
• Limit gt14.5 ps-1 _at_ 95 CL

• CDFII combined result
• Sensitivity 8.4 ps-1
• Limit Dms gt 7.9 ps-1 _at_ 95 CL

32
Future perspectives
33
Future perspectives

• Add more channels
• Bs?Ds3p (130 events 20)
• Bs?Dsp
• Add semileptonic Bs decays from the
• hadronic trigger (S. De Cecco talk)
• X2 semileptonic statistic
• Improve decay time resolution with PV event by
  event (detail)
• Incremental changes in existing algorithm (new
  Jet Charge 20 eD2)
• Add new tagging algorithm Same Side Kaon Tag
• New data rolling in, but increasingly peak
  luminosity
• Keep alive as much as possible present triggers
  ? SVT upgrade
• Use new trigger strategies
• 2 SVT Tracks opposite side muon (ptgt1.5 GeV) at
  trigger level
• (already in place since summer 2004 can survive
  at higher luminosity)

34
Same side Kaon tagging
Exploits the charge correlation between the b
quark flavour and the leading product of b
hadronization.
B0
Already used in Dmd measurement, gives an eD2
1.1 ? 0.4
p-K0
B case is complicated by the contribution of
excited Bd and Bs states
B-
pK

• SS Kaon tag possible with PID
• Issues
• Unlike opposite side tagger cannot calibrate
  using B0 and B
• Need to know eD2 from MC to set a limit on Dms
• MC tuning crucial

Bs
K- K0
35
MC-data comparison with PID
Apply PID, T.O.F. and dE/dx combined In a
Likelihood ratio L(K)/L(p)
Kaons

• Encouraging agreement!
• Issues
• Particle fractions in MC
• PID resolution tuning
• Backgrounds
• MC predict eD2 can be 2-3

Plong Kaons
36
Bsmixing sensitivity projection

• Analytic extrapolation, reproduce present result
  with current inputs
• Prediction include a reduced (50) effective
  luminosity usable for B-physics from 2007 onwards
• Sensitivity to the favorite CKM range
• In case of no signal 95 C.L. up to 30 ps-1 with
  4 fb-1
• CKM fit will imply New Physics if Dmsgt28 ps-1 by
  then

More projections
37
Summary

• First attempt at this (very) complex analysis!
• Expect close to 1fb-1 good data on tape by fall
  shutdown
• x3 statistics w.r.t to present result ?
• Additional channels can be used both for fully
  reconstructed and semileptonic
• Incremental improvements to existing opposite
  tagging algorithm expected
• Building confidence on Same Side Kaon tagging
• Better reconstruction of primary/secondary vertex
  improve proper time resolution
• Improved limit (15 ps-1 sensitivity?) expected by
  winter 06!
• Extensive upgrade to DAQ/trigger will keep
  B-triggers alive with increasing luminosity and
  allow the exploration of the SM favourite range
  for Dms by the end of RunII

38
Backup
39
The Upgraded CDF Detector
? Back
Central muon
Central calorimeters
Solenoid
New
Old
Partially new
Front end Trigger DAQ Offline
TOF
Endplug calorimeter
Silicon and drift chamber trackers
Forward muon
40
AD Projections (design plan)
you are here
41
CDF trigger architecture

• Crossing 396 ns 2.5 MHz
• Level 1 hardware
• Calorimeter, Muon, Track
• 25kHz (reduction x100)
• Level 2 hardware CPU
• Cal cluster, Silicon track
• 400 Hz (reduction x60)
• Level 3 Linux PC farm
• Offline quantities
• 90 Hz (reduction x5)

42
Basic tools PID

• Improved TOF calibration (better resolution)
• t0 (reduced tails)

TOF gt1? K/p separation up to p2 GeV

• Improved COT dE/dx calibration over wider bg range

dE/dx in COT K/p sep. gt1.4?_at_Ptgt2GeV
Combine TOFCOT in a likelihood ratio usable for
all momentum range!

• J/??ee, g?ee
• J/??mm
• D?D0 ps?(kp) ps
• L?pp

43
UT fit and Dms (CKMfitter)

• Yellow Band Dmd measurement 15 uncertainty
• Orange Band Lower limit on Dms Upper Limit on
  Vtd
• The lower limit on Dms already gives a
  constraint to the Triangle
• CKM Fit result Dms 17.86.7-1.6 (1s)
  15.2-2.7 (95CL)

from Dmd
back
Lower limit on Dms
from Dmd/Dms
44
CDF Bs Mixing Group

• 70 physicists ( 22 italians, 9 phd/post-doc) in
  CDF are actively involved in the Bs mixing
  project
• Improving the trigger strategy
• Understanding the detector
• B Lifetime Measurements
• Flavor Tagging
• B0 Mixing
• Bs Mixing
• Big collaborative effort
• Analyse 3 different datasets
• Reconstruct 0(20) different decay modes
• Perform 2 parallel analysis for both hadronic and
  semileptonic modes
• Study 4 different tagging algorithms
• TOF and dE/dx calibrations

45
Hadronic Bs CDF vs Aleph
30 events
500 events
46
Hadronic Bs signals (2)
NBS 25421
NBS 11618
back
47
Semileptonic Bs Signals (2)
157388 events
175083 events
back
48
Bs lifetime checks hadronic sample

• Raw lifetimes from mixing fit not good for
  averaging
• Average tB 1.5150.070 ps no systematics
  evaluated
• D0 t(Bs) 1.4200.0430.057 ps, WA t(Bs)
  1.469 0.059 ps

tB 1.3770.186
tB 1.5500.131
Ds?ppp
Ds?KK
49
Systematic Uncertainties
Hadronic
Semileptonic

• Dilution scale factors and templates
• systematic limited from control sample
• statistics

• Physics background at low Dms
• Prompt background at high Dms

Systematic errors are negligible with respect
to statistical in both cases
back
50
Systematics Summary Table (Hadronic)
51
Systematics Summary Table (Semileptonic)
52
Dilution scale factor error
53
CDF vs LEP
10
20
Dms ps-1
Dms ps-1
54
Likelihood Based Electron ID

• In CDF electron ID uses
• 10 parameters
• Calorimeter, tracking, dE/dx
• Preshower counter
• Shower max detectors

• Use likelihood to improve separation
• Typically improve the efficiency by 20 for
  the same purity

Fake
Electron
Combined Electron Likelihood
Electron
Fake
55
Electron Tag Performance

• For the Electron tagger, events are binnded in 2
  parameters
• Electron Likelihood
• pTrel
• Total eD2 (0.370.03)

• Sequential B decay
• b g cl-
• Higer pTrel
• b g c g sl
• Lower pTrel

56
Same Side tagging B0
Based on correlation between charge of
fragmentation p and flavor of b in B meson
Run II PRELIMINARY 2004
57
B0 mixing in the semileptonic channels
Muon Tag

• Measure Dmd
• Extract 5 dilution scale factors
• ? The dilution scale factors are used for
  semileptonic Bs mixing analysis

58
Other channels, example

• 133?23 Bs candidates
• Already used for lifetime
• But not for mixing
• 20 statistics

example
59
ct resolution improvements

• No EbE/L00
• s 67 fs
• With EbE/L00
• s 47 fs
• 30 improvement
• Not fully exploited yet (only L00)

back
60
NNet Jet Charge
Nnet based Wheight each track by its probability
to originate from b
?eD20.9 eD20.7 ?
61
SSKT MC tuning, no PID
One possible way to solve the issue of having a
prediction for the SSKT dilution is to extract it
from MC. ?Compare DATA with Pythia b-antib
production and hadronization with all the
processes on, underlying event tune A from HF
x-sec. CDF data.
Look at the charged tracks in a cone of DR0.7
around the Bs (no PID)
tracks
Plongitudinal
good agreement !
62
Bsmixing sensitivity projection(II)
back
63
Calibration B0 and B hadronic signals
64
Semileptonic B0 and B Signals
100K lD0
52K lD-
25K lD-

• B0 ?? B crosstalks
• B0 g lnD-
• B g lnD0
• with (D0 g D-p)

• Sample composition
• l D B0/B 85/15
• l D B0/B 85/15
• l D0 B0/B 20/80

65
Lifetime in the semileptonic Bs modes
ct 413.8 ? 20.1 mm
t 1.5210.040 ps
ct 422.6 ? 25.7 mm
Combined l-Ds lifetime result 445.0 ? 9.5 mm
(W.A. 438 ? 17 mm)
statistical err .only, ?NOT for
Averages? (DØ 05 426 ? 13 ? 17 mm)
Real Ds backgrounds prompt and physics
66
Experimental status on Dms
Present limit (HFAG 2004) from LEP / SLD / CDF
run I
Amplitude scan method discussed later

• 95 CL limit is Dms gt 14.5 ps-1
• Sensitivity 18.2 ps-1

67
Short term realistic scenario
Increase the actual effective statistics x4 (i.e.
increase NeD2 x4)

• Hadronic analysis will begin to lead the
  sensitivity
• Start to eat interesting Dms range combining
  the 2 analysis

68
Mixing Improvements

• Include Same Side (Kaon) Tagging
• Expect twice tagging power than OST combined
• x3 statistical power! but systematics limited
  in setting a limit
• Improve accuracy of primary vertex
• - 20 on s(ct) ?
• 40 on eD2 _at_ Dms 10 ps-1
• Add more channels 30
• Bs?Ds3p
• Bs?Dsp, Bs?Dsr
• x4 statistical power feasible with same data set
  ? x2 on amplitude error
• Sensitivity 15-16 directly from data
  extrapolation
• More statistics next year

CDF Run II combined
69
Decay Time Resolution

• Decay vertex error matrix overall correction for
  mis-knowledge of hit resolution
• ? Apply a scale factor S to s(ct) from vertex
  fit
• Huge control sample Ds random track to
  emulate Bs decay topology
• Correct for small (10) secondary Ds in the
  sample
• Parameterize S in terms of several variables (
  PT, Isolation,)
• Correct s(ct) Ss(ct) event by event.

Prompt track Ds vertex
Semileptonic Bs signal
Hadronic Bs signal
70
M(lD) Binning for K factor

• Resolution of K factor
• better for high M(lD)
• Dividing event in different M(lD)

• Evaluate K factor in each M(lD) bin
• Improve the decay time resolution

71
Prompt Background

• Clear prompt peak also in the right sign lepton
  D events
• Event tagged with high dilution tagger (Muon,
  Electron, Vertex)
• Prompt background is reduced
• No opposite side B for prompt BG

72
Tracking

• Great progress in Si stand-alone
• Substantial efficiency improvement at large h and
  low-pt
• Improved mass resolution
• L00 now ready for physics
• eff. 60 and growing
• Clear improvement in ?(Lxy),crucial for Bs mixing

NEW
73

• Strange peaks in expected Dms being there already
  before Tevatron new input

74
Flavor tagging Soft Leptons
Run I
Run II
Tag type eD2 ()
Muon (0.700.04)
Electron (0.370.03)
2ndary vtx (0.360.02)
Displaced track (0.360.03)
Highest p jet (0.150.01)
Total (exclusive) 1.6
75
Flavor tagging Jet Charge
Run II
Run I
Tag type eD2 ()
Muon (0.700.04)
Electron (0.370.03)
2ndary vtx (0.360.02)
Displaced track (0.360.03)
Highest p jet (0.150.01)
Total (exclusive) 1.6