Rare B Decays

Yee Bob HsiungNational Taiwan University

and Belle Collaboration

PASCOS 2006, Sept. 10-15, OSU, Columbus Ohio

Highlights from Belle and Babar

What is a rare decay?Experimental definition:One which has not been seen, or only just been

seen for the first time

Theoretical definition:One which, in the standard model, is either

absolutely forbidden or strongly suppressed:FCNC, helicity, small CKM element.

(b→c is big, so everything else is small)

What are rare B decays?

• Charmless B meson decays• Radiative decays• Leptonic decays• Charmless baryonic decays of B

Many, many rare B decay modes - BR, Acp, Pol, etc.

See later talks: CP Violation by Chunhui Chen and Elements of CKM Matrix by T. Tsumiyoshi

Why look for rare decays?

Our chance to see them iswhen the Standard

Model amplitudes aresmall:

==> Rare decaysu

bH+ τ

+

ν

If new particles are to appearon-shell at high energy

phenomenon, they mustappear in virtual loops and

affect amplitudes

Hadronic rare B decays also study SM, CP violation and NP.

What made this possible?

Measuring Branching ratios of~10-6 needs millions of events

1 fb-1 ~ 1M BB events!!

Physics progress possible thanksto machine physicists at SLACand KEK

• designed and built B factoriesoperating in a new high-current regime,

• continually faced andovercame new problems andchallenges.

• continually push theluminosities.

• Peak Lum ~ 16.5 /nb/s !!

KEKB/Belle 630 fb-1

PEPII/BaBar

-

Belle Apparatus and data set

e+(3.5GeV)(8.0GeV)e-(4S)

BB

KEKB

Data sample for analysisInt. L = 414fb-1 447MBB(Int. L=497fb-1 532MBB)

BaBar Detector

“Finding a needle in haystack.”Huge backgrounds from other Bs and e+e- →q q

Use energy balance ΔE=EB-∑Ei and beam-constraint

Mbc or MES=√EB2-(∑pi)2

“Blind Analysis”

Tune cuts without looking inthe Mbc-ΔE ‘signal box’

Use datasidebands

(rather thanMonte Carlo)to estimatebackground

MES(GeV/c2)

or

Mbc

ΔE(

GeV

)

Continuum Background suppression

Other Experimental techniquesContinuum

suppression fromcombined informationfrom shape variables

“Single-B beam” technique:

Reconstruct one “tag” B ina common decay mode

(hadronic or semileptonic).Remaining particles must

also form a B Limits fromN=ε S + b

N small

Uncertaintieson ε, b

B Decays to leptonsProceeds through one or two weak bosons with strongCKM suppression – door open for NP particles tocontribute

Free of hadronic uncertainties in final state - Clean

u

bH+ τ

+

ν

µ+

µ-

b

d,s

u

bW+ τ

+

ν

ν~

b

d,s W

W

u,c,tµ-

ν

µ+

Plusmanyother

diagrams

Rare B decay with “Missing Energy”

Motivation for B+→τ+ν

Β decay constantBF(B+τ+ ν) < 2.6 x 10-4 (BaBar)B. Aubert et al., PRD 73, 057101 (2006)

Most stringent published limit:

Sensitivity to newphysics fromcharged Higgs if theB decay constant isknown

Measuring Βτν is non-trivial

ϒ(4S)B- B+ντ

e+

ντ

νe

B+→τ+ντ, τ+→e+νeντ

B-→X

The experimental signature is rather difficult:B decays to a single charged track + nothing

Most of thesensitivity isfrom taumodes with1-prong

Belle’s sample of B tags (449 x 106 BB)

7 modes

6 modes0 (*) (*)1/ / / SB D a D! "

# + + + +$ +0 0

/D D! !" "

(*)0 (*)

1/ / / SB D a D! "+ + + + +# +

0 0 0/D D! " sD !

+

sD !+

0D !

D!"

sD

+! 2 modes

Beam constrained mass distn’s

Signal region : -0.08 < ΔE < 0.06 GeV, Mbc > 5.27 GeV/c2

~10% feed-acrossbetween B+ and B0

m ~ 5.28 GeV/c2

σ~ 3 MeV/c2 fromσ(Ebeam)

~ 180 channelsreconstructed

Charged B’s Neutral B’s

N=680 K

Eff=0.29%

Purity =57%

N=412 K

Eff=0.19%

Purity =52%

Outline of B τνexperimental analysis

• Reconstruct one B (Btag) in a charged hadronic b→ c mode (remove tag’s decay products fromconsideration.)

• Little or no extra electromagnetic calorimeterenergy (EECL) . Beam-related backgroundsmodeled in MC using random trigger data runs.

For B →X n known EB, mB, small pB– ⇒ narrow missing mass distn. (mn~0)

Two missing neutrinos, large missing p (cutdepends on τ decay mode 0.2 GeV-1.8 GeV)

Outline of experimental analysis (cont’d)• The τ lepton is identified in the 5 decay modes:

• Signal-side efficiency including τ decay BFs)

• All selection criteria were optimized before examiningthe signal region (a.k.a. blind analysis)

• Fit the extra energy distribution (EECL), the signal peaksnear zero

81% of all τ decays

15.81 ±0.05%

Consistency Check with BD* lν• Extra neutral energy EECL Validation with double tagged

sample (control sample);– Btag is fully reconstructed– Bsig is a semileptonic decay

BB++ D D(*)0(*)0 X X+ + (fully reconstruction)(fully reconstruction) B B-- D D*0 *0 l-ν DD00 ππ0

KK-- ππ++ K K-- ππ+ + ππ-- ππ++

502 ± 18Total458Data

7.9 ± 2.2B0B0494 ± 18B+B-

Purity ~ 90% Extra energy inthe calorimeter

Calibration data

Example of a Bτ ν candidate

Evidence for B+→ τν (Belle)

Find signal events from a fitto a sample of 54 events.4.6σ stat. significance w/osystematics,

449 ×106 B pairs Btag→D(*)[π,ρ,a1,Ds(*)] 680k tags, 55% pure.5 τ decay modes

5.3

4.717.2

+

!

MC studies show there is a small peakingbkg in the τππ0 ν and τπππ0 ν modes.

After including systematics(dominated by bkg), thesignificance decreases to3.5σ

Extra Calorimeter Energy

Error in the efficiency calculation

Due to a coding error, the efficiency quoted in the 1stBelle preliminary result was incorrect. The data plotsand event sample are unchanged. However, fB andthe branching fraction must be changed.

This mistake was not detected when checking the BD* l νcontrol sample or in the internal review process.

418.034.0

16.028.0 1006.1)(BF!++

!!

++"=# $%$B

Previousvalue

New value 439.056.046.049.0 10)79.1()(BF

!++

!!

++"=# $%$B (Preliminary)

Results for B+→τ+ν

439.056.0

46.049.0 10)79.1()(BF!++

!!

++"=# $%$B

Belle and BaBar results are similar.Agree within errors

Can be combined to give

(1.36 ± 0.48)x10-4

(new)BF(B+→τ+ντ)= (0.88 ±0.11) x10-4

BR< 1.80 10-4 @ 90%CL

(revised). 3.5 σ significance

+0.68-0.67

BF(B+→τ+ντ)

324M B’s

Direct experimental determination of fB• Product of B meson decay constant fB and

CKM matrix element |Vub|

• Using |Vub| = (4.39 ± 0.33)×10-3 from HFAG

fB = 216 ± 22 MeV (an unquenched lattice calc.)[HPQCD, Phys. Rev. Lett. 95, 212001 (2005) ]

15% 14% = 12%(exp.) + 8%(Vub)

36 30

31 34229Bf MeV

+ +

! != ( Belle)

1.6 1.1 4

1.4 1.3(10.1 ) 10B ubf V GeV+ + !

! !" = "

Constraints on the charged Higgs mass

rH=1.13±0.51

Assume fB and |Vub |are known, take theratio to the SM BF.

22 2

2(1 tan )B

H

H

mr

m!= "

Limits on e.g. 2Higgs doubletmodel: W.S.Hou,PRD 48, 2342(1993)

B± →µ ± ν and e ± ν

Helicity SuppressedUse hadronic tags: B fully

reconstructed as B to D(*) XLepton is monoenergetic in signal-B

rest frameLimits (@ 90% CL)

Motivation for BK*νν (bs with 2 neutrinos)

SM: BF(BK* νν) ~1.3 x 10-5 (Buchalla, Hiller, Isidori)

BSM: Newparticles inthe loop

c.f. SM: BF(BK- νν) ~4 x 10-6PRD 63, 014015

[Belle preliminary (277 x 106 B Bbar) : BF(BK- νν)

BK(*)νν are particularly interesting andchallenging modes (Bτν is even a small background)

The experimental signature is BK + Nothing

The “nothing” can also be light dark matter (mass oforder (1 GeV)) (see papers by M. Pospelov et al.)

(But need tooptimize pK cut)

DAMANaI 3σRegion

CDMS 04

CDMS 05

Direct dark-matter searches cannot see M

Search for BK*νν (535 x 106 B Bbar pairs)

0 *0 4( ) 3.4 10B B K !! "# < $ (at 90% C.L)

Extra Calorimeter Energy (GeV)

3.1

2.64.7Yield

+

!=

(1.7σ stat.significance)

Sideband = 19

MC expectation =18.7±3.3

SM (Buchalla, Hiller,Isidori) 1.3 x 10-5

BELLE-CONF-0627

Result from ablind analysis.

Search for BK*νν (properties of candidates)

KπInv. mass

b c background

rare B background (x 15 data)

udsc background

Data

combined backgroundSignal x 20

π+

K-

γ

Tag Side

B D+ a1-

D+ K- π+π+

a1- ρ0 π- , ρ0 π+π-

Event display for a BK*νν candidate due to an identified background (BK*γ)

(Hard photon is lost in the barrel-endcap calorimeter gap)

Missing mass ~ 0

MC: Expected bkg fromthis source ~0.3 evts.

B0 to l+ l- γ

FCNC and helicity suppressed, but aninitial state photon allows helicity flip

SM predictions of order 10-10(10-15, 10-11 respectively without the γ )See 0 events for e, 3 events for µ (but

compatible with background)Limits (at 90% CL)BR(B→eeγ)< 0.7 x 10-7BR(B→µµγ)< 3.4 x 10-7

(simulated)

Radiative B decays

FCNC process suppressed in SM: sensitive to newparticles in loops

b→sγ Inclusive and many exclusive measurements

b→sl+l -: More information from kinematics

b→dγ : strongly suppressed but open to different physics

b→d l+l -: on the way

B → s γ inclusive

‘Fully Inclusive’ and ‘sum of exclusives’ (38 modes)

NLO calculation (3.61 ) x10-4

result: (Eγ >1.9GeV) = (3.67 ±0.29 ± 0.34 ± 0.29) x10-4

HFAG average

(3.55 ± 0.24 ± 0.03) x10-4+0.09-0.10

+0.37-0.49

Branching Fraction now well measured. Theoryand experimental error similar

Not much room forNew Physics here

Constrains modelbuilders

B → s γ exclusive

• Lots of channels– Branching Fractions measured– CP violating asymmetries measured If nonzero these would be a signature of New Physics

Example: B→ K0sπ0γ

Δt(ps)

Belle: B → d γ

First Observation in 2005!

Combined BF

B → d γ

First observation of B+ → ρ+γ

B+→ρ+γ

B0→ρ0γ

MES(GeV/c2)

Compare with B→K*γ

b

d,s W

W

tb

t

d, s

tds

|Vtd/Vts|=0.171+0.018

-0.021

+0.017

-0.014

Same CKM elements as mixing –but a non-trivial test

Charmless Hadronic Decays• Many modes• Will present collected branching ratios• Will present measurements of time-integrated CP

violation ACP: they follow on directly from differencesin charge conjugate decay states– from the B+/B- difference - trivial– From self-tagged neutral modes – trivial– From C part of CP+mixing fit – nontrivial but standard

Observation of B to KK

MES(GeV/c2)Mbc(GeV/c2) ΔE(GeV)

2 body π-K combinations

B→ K+π- /K-π+

Direct CP violationExperiments agree:BaBar:ACP=-0.108 ±0.024±0.007

BelACP=-0.093 ±0.018±0.008

0B

0

B

Direct CP in K π

Competing amplitudes with different strong and weak phasesACP should be the same for K+π- and K+π0 (Gronau: hep-ph

0508047)

Current averages (HFAG)ACP (K+π-)=-0.093 ± 0.015ACP (K+π0)=+0.047 ± 0.026

Difference 0.14 ± 0.03 – a long way from zeroMaybe colour-suppressed trees are responsible

Maybe New Physics

B→K π ratios

Can form many ratios, especially(A Buras, R Fleischer et al, Phys J C 45 (701-710) 2006)

Rn=Γ(K+ π-) Rc=2 Γ(K+ π0) 2 Γ(K0π0) Γ(K0π+)

Obtain (HFAG averages)

Rn=0.99 ± 0.07

Rc=1.11 ± 0.07

Agree with each other

And with SM predictions

The “Kπ puzzle” is no more

B →ρ0ρ0

See 98 ±22 events

- 3 σ significanceBR (1.16 ±0.27)10-6

Fit longitudinal polarisation fl= 0.86 ± 0.05Measurement needed for B 0 →ρ+ρ- , used for alphaInforms penguin uncertainty in α determination

+32

-31

+0.36 -0.37

+0.11

-0.13

MES(GeV/c2) ΔE(GeV)

B+ → φ φ K+, B0 → φ φ K0

Signal Yield in 2D fit: 34.2 Significance: 9.5σ

B (B→φφK+): (3.18 ± 0.27) x10-6

AC P : 0.01 ± 0.02

+6.37- 5.77

+0.60- 0.52

+0.19- 0.16

Signal Yield in 2D fit: 7.27

Significance: 4.7σ

B (B→φφK0):(2.31 ± 0.24) x10-6

+3.04- 2.44

+1.00- 0.74

φφK0:

φφK+:

M

Threshold Enhancement

• PLB 617, 141-149, 2005

ppK+ ppK0S pΛπ-

Glueball is ruled out

Angular distribution: ppK+

bs dominant processFragmentation picture

p

p

K+ӨpX

Proton against K- (p against K

+) : flavor dependence!

ppK signal

Conclusion-I

• Tremendous progress in rare B decays for exploringSM parameters and probing physics at the TeV scale.

• Evidence for Bτν and experimental determination of fB(preliminary result has been updated)

• Search for BK* νν (UL is still a factor of 10 above theSM range)

439.056.0

46.049.0 10)79.1()(BF!++

!!

++"=# $%$B

0 *0 4( ) 3.4 10BF B K !! "# < $

Conclusion -II

• Limits on Higgs Masses, tan β, SUSYparticles.

• Many rare B decay modes are nowmeasured, many with upper limits toconstraint the SM parameters

• Direct CP violation in B to Kpi has been seen.• Standard Model begin to be stressed heavily,

new SuperB factory would stress it evenfurther with a potential for new discovery

Future Prospects: BτνΔfB(LQCD) = 5%

95.5%C.L. exclusion boundaries

tan /Hm!

tan /Hm!

rH

50ab -1 If Δ|Vub| = 0 & ΔfB = 0

4.4%3%50 ab-15.8%10%5 ab-17.5%36%414 fb-1Δ|Vub|ΔB(Bτν) expLum.

Extrapolations (T.Iijima)

Some modes are very difficult at hadron colliders

MC extrapolation to 50 ab−1

Observation of B± K± ν ν5σ

Super B LoI Fig.4.18

(compare to K+π+ννand KL π0νν)

Extra EM calorimeter energy

Belle result on Bτνshows that B to one prongdecays can be measured.

MC

SM pred: G. Buchalla, G. Hiller,

G. Isidori (PRD 63 014015 )

Backups

Bτν yields broken down by τ decay mode

For all modes, the background is fitted with a 2nd orderpolynomial plus a small Gaussian peaking component.

(stat sigonly)

Fits to individual Bτ ν decay modes(updated for ICHEP06)

P*_K*

b c background

rare B background (x 15 data set)

udsc background

Data

combined backgroundSignal shape

Search for BK*νν (properties of candidates)

K* momentum distribution

Need more bcMC (only 2 x data)

Lipkin Sum Rule

RLipkin=2 Γ(B+→K+π0)+Γ(B0→K0π0)

Γ(B+→K0π+)+Γ(B0→K+π-)From isospin and assuming the b →s penguin diagram

dominates R should be 1+O(10-2) Obtain (HFAG average)

RLipkin=1.06 ± 0.05(Was 1.25 ± 0.10 in 2003)