1. Measurements of Radiative Penguin B Decays at BaBarTM
Measurements of Radiative Penguin B Decays at BaBar
All results preliminary unless explicitly indicated
Jeffrey Berryhill
University of California, Santa Barbara
For the BaBar Collaboration
32nd International Conference on High Energy Physics
August 17, 2004

2. Radiative Penguin Decays and New Physics
Radiative penguin decays: b → sg and b → dg FCNC transitions
SM leading order = one EW loop
Vts, Vtd dependent
FCNCs probe a high virtual energy scale comparable to high-energy colliders
Radiative FCNCs have precise SM predictions:
BF(b→sg)TH = 3.57 ± 0.30 x 10-4 (SM NLO)
BF(b→sg)EXP = 3.54 ± 0.30 x 10-4 (HFAG)
Decay rate agreement highly constrains new physics at the electroweak scale!
Further tests presented here:
Exclusive b→sg decay rates
b→sg CP asymmetries
b→dg penguins
Multiple new BF(b→sg)
measurements coming soon from BaBar
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

3. Measuring Exclusive Radiative Penguin Decays
Small rates (BF < 10-4) on top of large backgrounds
Common strategy :
select photon + hadrons with strict particle ID
cut when possible on meson masses
reduce continuum background with multivariate methods
extract signal with multi-dimensional maximum likelihood fit
Ex: signal fit for B → K*0g, K*0 → K+p-
Best kinematic constraints
from B candidate momentum and
energy, compared with Ebeam:
583±30 events
* denotes CM frame
For signal, mES ≈ mB, s(mES) ≈ 3 MeV
DE* ≈ 0, s(DE*) ≈ 50 MeV
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

4. B  K* g Branching Fractions: Summary
BaBar preliminary measurements on 82 fb-1
Becoming systematics limited
Exp. vs. theory: data more accurate
than form factor predictions!
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

5. Direct CP Asymmetry: b →sg and B  K* g
< 1% in the SM, could receive ~10% contributions from new EW physics
Either inclusive or exclusive decays could reveal new physics
B or K charge tags the flavor of the b quark with ~1-2% asymmetry systematic
Sum of 12 exclusive,
self-tagging
B→Xs g final states
Xs = K/Ks pions
Eg* > 2.14 GeV
b→sg
b →sg ACP = (N – N)/(N + N) = ± ± 0.015
PRL 93 (2004) , hep-ex/
Asymmetries also measured precisely in exclusive K*g decays:
B →K*g ACP = ± ± 0.010
submitted to PRL, hep-ex/
D0- = G(K*0 g) – G(K*- g) = ± ± ± 0.024
G(K*0 g) + G(K*- g)
preliminary
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

6. b  s g Asymmetries: Summary
BaBar measurements on 82 fb-1
K*g, K2*g preliminary; Xsg published
CP asymmetries consistent with SM (0.4%) at the ~5% level
K*g isospin asymmetry D0- consistent with C7 < 0
Statistics limited up to ~1 ab-1
sgn D0- = -sgn C7
Can we exclude C7 > 0 ?
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

7. Time-Dependent CP Asymmetry in B →K*g (113 fb-1)
As in B0→J/y KS, interference between mixed and
non-mixed decay to same final state required for CPV.
In the SM, mixed decay to K*g requires wrong photon
helicity, thus CPV is suppressed:
In SM: C = -ACP ≈ -1% S ≈ 2(ms/mb)sin 2b ≈ 4%
Measuring Dt of K*(→KSp0) g events requires novel beam-constrained vertexing techinque:
Vertex signal B with intersection of KS trajectory and beam-line
Usable resolution for KS decaying inside the silicon tracker
Validated with B0→J/y KS events
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

8. Time-Dependent CP Asymmetry in B →K*g (113 fb-1)
Likelihood fit of three components
(qq, BB, K*g)
to 5D data
(mES,DE,Fisher,mK*,Dt)
K*g signal = 105 ± 14 events
S = ± 0.63 ± 0.14
C = ± 0.32 ± 0.09
submitted to PRL, hep-ex/
Consistent with SM
For C fixed to 0, S = 0.25 ± 0.65 ± 0.14
preliminary
First ever measurement of time-dependent CP asymmetries in radiative penguins!
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

9. Search for B → rg, wg (191 fb-1)
NEW!
Simplest and most “common” b → dg exclusive decays are rg and wg:
B+ → r+ g, r+ → p+ p BF ≈ BF K*g x |Vtd/Vts|2 ≈ 1.6 x 10-6
B0 → r0 g, r0 → p+ p BF ≈ ½ BF r+g
B0 → w g, w → p+ p- p BF ≈ ½ BF r+g
Previous BaBar limit: BF r+g < 1.9 x % CL (78 fb-1)
Preliminary Belle evidence (3.5s): BF r+g = 1.8 ± 0.6 ± 0.1 x 10-6 (140 fb-1)
Comparable to rarest
B decays measured!
Ratio of exclusive b → dg and b → sg decay rates measures |Vtd/Vts| via
Ali et al. hep-ph/
DR weak annhilation correction
z2 SU(3) symmetry breaking of exclusive form factors
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

10. Continuum Background Suppression
e+e- → qq continuum events are
the dominant background
Combine properties discriminating
B → g + X from continuum
into a single-output neural net:
Event shape variables: spherical B vs.
jet-like continuum
B flavor tagging variables: kaons and
leptons in the rest of the event
B candidate vertex separation from
rest of the event
Neural net trained on simulated signal
and continuum samples
e = 76%
cut
Neural net distribution for both signal and background described well by simulation
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

11. B Background Suppression
Large (40 x signal) K* g background
reduced by strict p+ PID requirements
(track dE/dx, DIRC qc , DIRC Ng )
K+ misid 1-2% for most of acceptance
DE provides additional separation
of K*g from rg
B decays to rp0, rh, wp0, wh
suppressed by cut on helicity angle
Combine helicity angle, B production angle,
Dalitz angle (w only) into B Fisher discriminant
for additional background separation in signal fit
cut
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

12. Signal Fit: K*g Control Sample
Validate fit procedure on
data with K*g sample
(r0g candidates with
no PID requirement)
Unbinned extended ML
fit of three background components
(qq, Xs g, r0(p0/h))
and K*g component
to 4D data (mES,DE,NN,Fisher)
qq, Xs g, K*g yield are free parameters
Large K*g signal yield
consistent with BaBar
BF measurement
K*g + background
preliminary
DE shifted
background
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

13. Jeffrey Berryhill (UCSB)
Signal Fit: B+ → r+ g
r+g + background
N(signal) = 26 ± 15 ± 2 significance = 1.9 s
e = 13.2 ± 1.4 %
BF = (0.9 ± 0.6 ± 0.1) x10-6
BF < 1.8 x % CL
preliminary
background
Fit of four backgrounds
(qq, Xs g, r+(p0/h), K*g)
and r+g signal
qq, Xs g, r+g yield are free parameters
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

14. Jeffrey Berryhill (UCSB)
Signal Fit: B0 → r0 g
small peaking background
N(signal) = 0.3
significance = 0.0 s
e = 15.8 ± 1.9 %
BF = (0.0 ± 0.2 ± 0.1) x10-6
BF < 0.4 x % CL
preliminary
Fit of four backgrounds
(qq, Xs g, r0(p0/h), K*g)
and r0g signal
qq, Xs g, r0g yield are free parameters
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

15. Jeffrey Berryhill (UCSB)
Signal Fit: B0 → w g
- 4.5 – 1.9
N(signal) = 8.3
significance = 1.5 s
e = 8.6 ± 0.9 %
BF = (0.5 ± 0.3 ± 0.1) x10-6
BF < 1.0 x % CL
preliminary
Fit of two backgrounds
(qq, w(p0/h))
and wg signal
qq, wg yield are free parameters
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

16. Jeffrey Berryhill (UCSB)
Combined Fit: B → r, w g
Simultaneous fit to all three samples with the constraint
Combined BF(r,wg) ≡ BF(r+g) =
2(t+/t0) BF(r0 g) =
2(t+/t0) BF(w g)
BF = (0.6 ± 0.3 ± 0.1) x10-6
significance = 2.1s
BF < 1.2 x % CL
preliminary
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

17. b  d g Branching Fractions: Summary
Comparison of measurements with predictions for individual modes and combined BF
central value
90% C.L. upper limit
2.0s difference in
BaBar, Belle BF
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

18. Jeffrey Berryhill (UCSB)
CKM matrix constraint
BaBar BF ratio upper limit < → |Vtd/Vts| < (90% CL)
Ali et al. hep-ph/
(z2,DR) = (0.85,0.10)
no theory error
Penguins are starting to
provide meaningful CKM
constraint
rg 95% C.L. BaBar allowed region (inside the blue arc)
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

19. Jeffrey Berryhill (UCSB)
CKM matrix constraint
SU(3) breaking of
form factors z2= 0.85 ± 0.10
weak annhilation
correction DR = 0.1 ± 0.1
Ali et al. hep-ph/
(z2,DR) = (0.75,0.00)
(z2,DR) = (0.85,0.10)
theory error
no theory error
Penguins are starting to
provide meaningful CKM
constraint
Reduction of theory errors
necessary to be competitive
with Bd,Bs mixing
rg 95% C.L. BaBar allowed region (inside the blue arc)
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

20. Jeffrey Berryhill (UCSB)
Summary
Decay rate measurements of b→ sg penguins are well into the precision era.
CP asymmetries of b→ sg penguins are statistics limited and will continue to test
the SM
ACP in B → Xs g PRL 93 (2004) , hep-ex/
BF, ACP, and Isospin asymmetry in B→ K*g submitted to PRL, hep-ex/
time-dependent CPV in B0 → K*0 g submitted to PRL, hep-ex/
b→ dg penguins are only now beginning to reveal themselves in B-factory data.
They could also uncover new physics, or measure the poorly known |Vtd|.
BaBar finds no evidence for B → rg in 211 million BB events
submitted to PRL, hep-ex/
BF(B → r,w g) < 1.2 X |Vtd/Vts| < (90% CL)
First measurement!
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

21. TM
Backup Slides

22. B → K*g (82 fb-1): Signal fits
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

23. B → K*g (82 fb-1): K*g lineshape
K* lineshape for
events with:
mES > 5.27 GeV/c2
-0.2 GeV < DE < 0.1 GeV
Background subtracted
with shape from mES
sideband
Fit to relativistic
Breit-Wigner with
PDG G(K*), mK*
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

24. Time-Dependent CP Asymmetry in B →K*g
Z resolution vs. KS decay radius
Projection of fit with likelihood cut
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

25. B → rg: Neural net performance
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)

26. B → rg: K*g Control Sample
r+g, no PID
r0g, no PID
ICHEP 17 Aug 04
Jeffrey Berryhill (UCSB)