1. 1 Search for light Higgs in Y(1S)→ gamma lepton-pairs Nasra Sultana & Tomasz Skwarnicki

2. 2 Motivation Some NMSSM models (Dermisek, Gunion, McElrath: hep- ph/0612031) predict existence of a new non-SM-like higgs boson a 0 (pseudo-scalar) with m a < 2m b to avoid fine-tuning of parameters in electroweak symmetry breaking Such light higgs avoids the LEP limit m H > 100GeV based on e + e - →ZH(→bb) searches since its mass is below the threshold for decay to bb. In this scenario also SM-like higgs boson h (scalar) also avoids the LEP lower mass limit since Br(h→ bb) is much smaller than Br(h →a 0 a 0 ) The perfect place to search for a 0 is in radiative decays of Upsilon meson, Υ →  a 0. Such an a 0 decays predominantly into heaviest pair of fermions available (Br(a 0 →     )~0.9 for m a >2m  ) We have studied the decay Υ →  a 0 followed by a 0 →     (or a 0 →     for m a <2m  

3. 3 Previous results are very old ARGUS: Phys.Lett.B154:452,1985 Υ(2S) →     Υ(1S), Υ(1S)→     ) CUSB: Phys.Rev.D35:2883,1987 Υ(1S)→  X

4. 4 Method (a 0 →     ) To study the decay Υ (1S)→  a 0, a 0 →      we tag Υ(1S) via Υ(2S) →     Υ(1S). By tagging the Υ(1S) we eliminate events with photon coming from initial state radiation in tau pair production (e+e-→       ), a serious background for the reactions e + e - → Υ(1S)→  a 0. The channel a 0 →      is selected by using 1- prong  decays, requiring missing energy (neutrinos!) and at least one leptonic  decay:  →  or  →e 

5. 5 Cuts details Numbers of charged tracks = 4 -0.015 < Recoil-Mass(  ) – M(Y(1S)) < +0.015 GeV Require at least one of the remaining 2 charged tracks to be an electron or muon candidate: – e  E/P-1 | < 0.15, DEDX:  e  –  : depthmu >1, muqual=0, 0.15< E< 0.45, DEDX:    Select the highest energy photon in the good barrel part (E  > 0.06 GeV) which does not make a   mass within 3  with any other photon as a candidate for  Υ(1S) →  a. The  0 veto suppress  → ,  →     →  background Sum up energy of all other photon candidates: E neutral Imbalance of total energy: E  + E charged + E neutral – E cm < -0.5 GeV Mass of neutrals (except for the highest energy  ) plus the 1- prong not required to be a lepton < 2 GeV cos(1-prong and  )< 0.99 to suppress final state radiation.

6. 6     Recoil mass – Υ(1S) mass Signal region Side band

7. 7 9.4 M Υ(2S) decays Photon Energy distribution in the rest frame of Υ(1S) Scaled side bands (non Y(1S) background)

8. 8 Photon Energy distribution in the rest frame of Υ(1S) after side band subtraction e + e - →     Υ(1S), Υ(1S)→ l  l  MCs scaled by PDG BRs Sideband-subtracted data Data above 200 MeV saturated by e + e - →     Υ(1S),Υ(1S)→     Within errors all data well described by Υ(1S)→ l  l  We used Υ(1S)→     MC to optimize our data selection procedure. Υ(1S)→     MC + →     MC + → e  e  MC

9. 9 Photon Energy distribution for various m a e + e - →     Υ(1S) Υ (1S)→  a 0, a 0 →     signal MonteCarlo 10,000 events for each mass BKH’s fix to MC energy resolution is on Peaks are fitted with a Crystal Ball function Signal MC: m a = 9 GeV m a = 8 GeV m a = 7 GeV m a = 6 GeV m a = 5 GeV m a = 4 GeV m a = 8.5 GeV m a = 9.15 GeV m a = 9.30 GeV m a = 9.35 GeV m a = 9.41 GeV

10. 10 Efficiency obtained from fits to signal MC and interpolated for the regions in between. Fits to MC data (previous slide) Polynomial fit to interpolate to other photon energies (used in calculation of upper limits on signal BR) Plotted efficiencies based on phase-space MC Multiply them by 0.91 to account for 1+cos 2 θ  distribution for Υ→  a

11. 11 Energy resolution obtained from fits to signal MC (points) and interpolation to other energies (solid line). Obtained by BKH and Selina (CBX 02-22) from fits to single  MC (before the MC resolution fix) Factor from fits to our MC

12. 12 Photon spectrum with binning comparable to expected signal width

13. 13 Setting upper limits on signal yield At each energy fit CB line shape with width determined from MC on top of linear background in the ±  ln(E) = 0.5 range around the peak Fix signal amplitude at values  minimize with respect to the background parameters, then plot the fit likelihood as a function of the signal amplitude Determine 90% U.L. on the signal amplitude by integral of the likelihood function Example for ln(E in MeV)=7.5 90%

14. 14 Systematic errors ContributionValue Neglected     angular correlations and helicity correlations in     decays 20 % Track reconstruction (per event) 4 % Photon detection 3 % Number of Y(2S) decays 1.5% Error on BR(Υ(2S)→     Y(1S)) 3.2% Total systematic error 21 %

15. 15 Upper Limit on product branching ratio Br(Υ(1S)→  a 0 )*Br(a 0 →     ) as function of m a Br(Υ(1S)→  a 0 )*Br(a 0 →  ) = Ns / (  * N Υ(2S) * Br(Υ(2S) →  Υ(1S) ) Br(Υ(2S) →  Υ(1S))=18.8 % PDG’06 Upper limits are loose at low photon energies (E  <150 MeV) since our analysis was optimized for intermediate and high energies.

16. 16 CLEO III We have improved ULs by about an order of magnitude or more. We are constraining NMSSM models. Many models with 2m  <m a <7.5 GeV (represented by red points) ruled out by our results. Switch to a 0 →     for m a <2m  (blue points) - see next! From Dermisek, Gunion, McElrath: hep-ph/0612031 NMSSM consistent with all previous results

17. 17 a 0 →     Two identified muons: – depthmu >1, muqual=0, 0.15< E< 0.45, DEDX:    – RICH:  2 K1  2   2 K2  2   0 | E  + E charged – E cm |< 0.25 GeV e + e - →     Υ(1S), Υ(1S)→     MC scaled by PDG BRs ( includes tiny Υ(1S)→     contribution) Data (sideband subtraction very small)

18. 18 a 0 →     m a Gev Efficiency (  )% 0.35.6±0.2 0.55.4±0.2 1.05.6±0.2 2.05.4±0.2 3.06.1±0.2 Data Signal MC m a =3 GeV 0.5 GeV

19. 19 Using  =5.4% Br(Υ(1S)→  a 0 )*Br(a 0 →     ) < 2.5 x 10 -5 (90% C.L.) Eliminates most of NMSSM models for m a < 2m  (blue points) Concerns about ability of our MC to correctly predict tracking efficiency for very small m a (no opening angle between tracks) Do not intend to show any a 0 →  +  - results in public at this point

20. 20 a 0 →     Using  =4.3% Br(Υ(1S)→  a 0 )*Br(a 0 →     ) < 3.2 x 10 -5 (90% C.L.) Same concerns about MC as for     m a GeV Efficiency (  ) % 1.04.3±0.2 2.07.6±0.3 3.06.9±0.3 4.06.6±0.2 Data Signal MC m a =4 GeV 2.0 GeV Two identified kaons: –RICH:  2 K  2   0 –DEDX:    –depthmu <1 | E  + E charged – E cm | < 0.25 GeV

21. 21 Summary and plans We have obtained meaningful limits on Br(Υ(1S)→  a 0 )*Br(a→     ) and *Br(a 0 →     ) Future work: –Study effects of angular correlations in MC to reduce systematic error –Try separate set of cuts to optimize for high a 0 masses (E  < 150 MeV) ? –Study track reconstruction efficiency for low a 0 masses in a 0 →     with e + e - →  followed by  conversion (  →e + e - ) –David McKeen and Jon Rosner performed theoretical calculations which indicated that direct Y(1S) production (e + e - → Y(1S)) will be more effective than Y(2S) →     Y(1S) in setting limits for low mass a 0 →    . We will investigate this with CLEO data and MC.

22. 22 CLEO III We have improved ULs by about an order of magnitude or more. We are constraining NMSSM models. Many models with 2m  <m a <7.5 GeV (represented by red points) ruled out by our results. Switch to a→     for m a <2m  (blue points) - see next! From Dermisek, Gunion, McElrath: hep-ph/0612031 NMSSM consistent with all previous results