Steven Blusk, Syracuse UniversityRecontres de Moriond, March 2005

1
Measurements of Hadronic, Semileptonic and
Leptonic Decays of D Mesons at Ecm3.77 GeV in
CLEO
Steven Blusk Syracuse University
CLEO-c

• Outline
• Introduction
• Hadronic Branching Fractions
• Semileptonic Decays
• D?m nm
• Conclusion

2
CESR-c/CLEO-c

• The CLEO program has migrated from running on the
  ? resonances to the region around the y(3770)?
  Charm factory to study D, Ds mesons -- Broad
  program of charm physics (3 fb-1 goal)
• Additional running at J/y (search for exotics in
  radiative decay) in year 3

CLEO-c

• Tracking (93 of 4p)
• 16 axial, 31 stereo layers
• sp/p 0.6
• CsI (93 of 4p)
• 6144 crystals (barrel only)
• sE/E 5 at 100 MeV 2.2 at 1
  GeV
• Particle ID
• RICH (80 of 4p) dE/dx
• eKgt90 for p fakelt5

Inner DriftChamber
Accelerator changes - installation of SC
wigglersto improve damping ? higher L

• CLEO-c detector largely same as CLEO-III,
• Silicon replaced with drift inner chamber
• B field reduced from 1.5T?1.0T

Log
y
y(3770)
Ebeam
3
Some highlights of the CLEO-c Charm Program

• Precision measurements of D branching fractions
• Precise measurements of fD and fDs, the D decay
  constants.
• When combined with LQCD will enable 5
  determinations of Vtd and Vts
• Pave the road for a more accurate extraction of
  Vub
• Measurements of D?pln and D?rln form factors
  will provide testedlattice QCD predictions on
  heavy-to-light FFs.
• Extraction of Vcd, Vcs
• Unitarity Triangle
• Once Vub Vtd are measured to O(5) ?
  Allows for a stringent test of CKM angles (ie.,
  sin2b) vs sides

4
CLEO D Tagging

• Pure DD final state, no additional particles (ED
  Ebeam).
• Low particle multiplicity 5-6 charged
  particles/event
• Good coverage to reconstruct n in semileptonic
  decays
• Pure JPC 1- - initial state

• Tag one D meson in a selected tag mode.
• Dictates whether final state is DD- or D0D0
• Study decays of other D, (signal D)

K
Dsig
e
e-
Dtag
p
p
ED? Ebeam improves mass resolution by 10X
Analysis Preview

• Hadronic BF Use double-tagged and single-tagged
  yields
• Semileptonic decays Dtag (Dsig ?Xene),
  reconstruct ne using Pmiss
• Leptonic Decays Dtag (Dsig ?mnm)

Analyses shown today based on 57 pb-1
5
Absolute D Hadronic Branching Fractions
Single Tags Reconstructed one D meson Double
Tags Reconstruct both D mesons
D0 ? K-p D0 ? K-pp0 D0 ? K-ppp-
D ? K-pp D ? K-p pp0 D ? Ksp D ?
Kspp0 D ? Ksppp- D ? K-Kp
D- ? Kp-p- D- ? Kp- p-p0 D- ? Ksp- D- ?
Ksp-p0 D- ? Ksp-p-p D- ? K-Kp-
6
Fits to Data
D Modes
D0 Modes

• Signal shape includes
• y(3770) line shape, ISR, beam energy spread
  momentum resolution

Kp
Kspp0
Kpp
Ksppp
Kpp0
Kppp0
Ksp
KKp
Efficiency includes FSR losses
Kppp
7
Systematic Uncertainties
Tracking, p0 and Ks all use similar missing
mass technique.

• For pion
• Look at mass recoiling against J/yp in
  y?J/ypp- events ? Peak at Mp2 for
  J/ypp-.
• Count the number of times the track is found
  versus not found.

p track found
p track not found
MC
DATA
Uncertainty ? 0.7 / (p/K)
8
Preliminary Results
to be submitted to PRL
D Modes
D0 Modes
Normalized to PDG
As many of the systematics are evaluated using
data, they will shrink as ?L
9
Semileptonic Decays
?e
Vcs , Vcd
e
W
c

• Test LQCD on shape of f(q2) ?Use tested
  Lattice for norm.
• ?From B(D?Xen) extract Vcd
• D?p FF related to B?p FF by HQS ? Precise
  D?p FFs can lead to reduced stheory
  in Vub at B factories
• Similar for D?Vln, except 3 FFs enter

• Can also form ratios, where theory should be
  more precise

LQCD, PRL 94, 011601 (2005)
10
Basic Technique
D0 Tag Modes
MBC
11
Pseudoscalar Modes D ? Pene
c?s Cabibbo Favored
c?d Cabibbo Suppressed
(110 events)
(1400 events)
Events / ( 10 MeV )
Events / ( 10 MeV )
U Emiss Pmiss (GeV)
U Emiss Pmiss (GeV)
(60 events)
(500 events)
Events / ( 10 MeV )
Events / ( 10 MeV )
U Emiss Pmiss (GeV)
U Emiss Pmiss (GeV)
12
Vector Modes D ? Vene
57 pb-1 Data
c?s Cabibbo Favored
c?d Cabibbo Suppressed
First Observation
First Observations
(30 events)
(90 events)
U Emiss Pmiss (GeV)
(8 events)
(5s)
(30 events)
(400 events)
U Emiss Pmiss (GeV)
13
Preliminary Results
to be submitted to PRL
14
Inclusive SemileptonicElectron Spectra
CLEO-c Preliminary
D0 D
15
Leptonic Decay
G(B?tn) 10-4 10-5 difficult
Goal Extract fD, and eventually fDs (with
precision) ? Test LQCD, if it passes then trust
it in predicting fB, fBs ? Critical to
measuring Vtd/Vts, one of the sides of the UT
16
The Technique
17
Results
CLEO-c Yellow Book 1 fb-1
MostlyKLpbackground
8 events
DATA
18
Summary

• CLEO-c is off to a great start.
• With only 57 pb-1 on y(3770) (3 fb-1 proposed),
  measurements are already comparable or better
  than world average.
• Many more analyses are in the pipeline which I
  havent had timeto discuss.
• Many more exclusive BRs being investigated
• Several variants of inclusive and exclusive SL
  analyses
• Techniques for estimating systematics
  established using data.
• With more data, they will be reduced.
• Look forward to many precision results in charm
  physics comingfrom CLEO.

19
Backup Slides
20
Particle ID

• Use modes where the particle content is
  unambiguous.
• For p D?K-pp, D0 ?Kspp, D0 ?Kpp0
• For K D?K-pp, D0 ?Kpp0
• Then apply tagging requirements
• If both p K hypotheses analyzed,
• 3s dE/dx consistency, and
• D ( s(dE/dx)p2 - s(dE/dx)K2 LogLik(p)-
  LogLik(K)lt 0 (Nggt2)
• Drop RICH if
• RICHDONE is false, or, p(p)lt0.55 GeV/c, or
  cosqgt0.8

Correction applied (0.30.3) for p and
(1.31.3) for K
21
D Hadronic Systematics

• DE requirement compare yields with without DE
  cut (1.5)
• FSR Validated using J/y?mm, conservatively 0.5
  for ST, 1 for DT
• G3770 Lowe from 30.6 MeV ?23 MeV and take shift
  in data as systematic (0.6)
• Resonant substructure affects efficiencies,
  depending on mode 0.4 1.5
• Trigger efficiency trigger simulation ?
  0.1(Ksp) 0.2 (K-pp0)
• Multiple candidates
• Multiple candidates can result in choosing the
  wrong combination resultingin a loss in
  efficiency.
• MC does not model the number of multiple
  candidates/event well.
• Affects modes with p0s
• 1.30 for Kspp0, 0.44 for K-p pp0, 0.32
  for K-pp0
• Double DCSD Unknown relative phase between DCSD
  CAD amplitudes (0.8)
• Fit functions 0.5
• Data processing 0.3
• Quantum (CP) correlations Negligible

22
Yield Extraction in D Hadronic
CBX 05-06, A. Ryd.

• Signals are fit using
• y(3770) line shape, ISR, beam energy spread,
  momentum resolution(G3770 set to 30.6 MeV, as
  determined from data reduced to WA for
  systematics)
• Fit double tags first, using D?X / D?X, then fit
  single tags with sig pars fixed
• Disentangle momentum resolution from beam energy
  spread

• Signal Resolution (Signal MC) (5
  parameters per D)
• 3 Gaussian widths (s1,s2,s3) 1.5s1lt s2lt 4s1
  1.5s2lt s3lt 4s2
• Two fractions f2, f3, (1-f2-f3)
• Fixes resolution for double single tag fits in
  MC data

Beam Energy, inc. ISR
Signalregion
All candidates

• Background
• 1 correct D 1 incorrect D (fsig
  ARGUS)
• Mispartitioning of daughters
  ARGUS(ltMgt)GAUSS(DM)
• Both Ds are background ARGUS(MD1)ARGUS(MD2)

Kpp0
Kpp0
Double TagFit to Signal MC
23
Branching Fraction Fitting
CBX 04-36 (W. Sun)
Corrected yields are given by

• n Raw yields of single double tags
• b estimated backgrounds from other D modes

• N Fitted yields of single double tags
  NDDBi
• E Efficiency matrix
• diagonal elements are efficiencies
• off-diagonal are cross-feed probabilities
• F background probability matrix

Test using Toy MC- 3 neutral 2 charged modes
- no biases- proper error estimation
V is the variance matrix, and contains both
statistical systematic uncertainties
Since eij ? ei ej, correlated systematics cancel
in NDD
To first order, Bi is independent of tag modes
efficiencies.
24
Backgrounds
Single Tags
Cross-feed
External Not simulated in MC
Double Tags assume only 1 fake contributes,
since P(2 fakes) very small
Backgrounds that are dependent on fit
parameters, ie., NDD, are updated after each
iteration..
25
Backgrounds - II
MBC distributions for generic MC after signal
modes and backgrounds considered are removed
MBC distributions for non-DD MC
26
Fit to Generic MC(50X Data!)
Worst difference is 2.1s, for Kpp But this is for
50X data ? Scale by ?50 for data ?
0.3sstat. Deemed acceptable by committee, and
noted in PRL.
27
Form Factor Shapes
No efficiency corrections, resolution 0.025 GeV2
Future goal slopes 4, form factors over all q2