1. Search for Thermal Photons in PHENIX - Torsten Dahms - Stony Brook University 23 rd Winter Workshop On Nuclear Dynamics February 13, 2007

2. Torsten Dahms - Stony Brook University2 Outline Motivation: signature of QGP Conventional direct photon measurements New attempts at low p T : –Tagging of EMCal photons –External conversions in detector material –Low mass internal conversions Summary

3. Torsten Dahms - Stony Brook University3 The “Little Bang” in the lab  space time Hard Scattering Au  Expansion  Hadronization Freeze-out QGP Thermalization electro-magnetic radiation: γ, e + e -,  +  - rare, emitted “any time”; reach detector unperturbed by strong final state interaction Final state probesSoft probesPenetrating probes e- e+ 

4. Torsten Dahms - Stony Brook University4 Thermal photons? hard: thermal: Decay photons (  0 → ,  → , …) No significant excess at low p T

5. Torsten Dahms - Stony Brook University5 Direct photons in Au+Au Conventional method: Measure inclusive photons γ incl = γ decay + γ direct Calculate double ratio: (γ incl /π 0 ) measured / (γ decay / π 0 ) background = γ incl / γ decay =1+ γ direct / γ decay If double ratio > 1  direct photons Run4: more statistics, but still no conclusive measurement

6. Torsten Dahms - Stony Brook University6 The PHENIX experiment Charged particle tracking: –DC, PC1, PC2, PC3 Electron ID: –Cherenkov light RICH –shower EMCal Photon ID: –shower EMCal Lead scintillator calorimeter (PbSc) Lead glass calorimeter (PbGl) –charged particle veto e+e+ ee  

7. Torsten Dahms - Stony Brook University7 New Idea Measure inclusive photon spectrum, but a very clean sample Tag all photons, which combined with a photon from a second (less clean) sample can be identified as pion decay product DATA a  =  acceptance   =  efficiency   ’ = loose  efficiency f = conditional probability of having a loose photon (  ’) in the acceptance, once you already have a clean  the acceptance SIMULATION DOUBLE RATIO everything cancels out except for   ’     minimal systematics

8. Torsten Dahms - Stony Brook University8 Clean Photon Sample Method I: –Only use EMCal clusters which fulfill very strict PID cuts Method II: –Identify conversion photons in beam pipe –Additional advantage: very good momentum resolution of charged tracks at low p T No detector artifacts no pair cut with pair cut Dalitz Conversions r~mass PHENIX Beam Pipe 

9. Torsten Dahms - Stony Brook University9  0 signal extraction Real events Mixed event BG subtraction within p T bins Normalized outside the π 0 peak Clean EMCal sample has better S/B ratio Clean EMCal γ sample Beam pipe conversions

10. Torsten Dahms - Stony Brook University10 Results in Au+Au Agreement of all three results within their errors There seems to be an excess above the decay photons at low p T

11. Torsten Dahms - Stony Brook University11 Compton q  g q q  g q e+e+ e-e- phase space factorform factor invariant mass of virtual photon invariant mass of Dalitz pair phase space factorform factor invariant mass of Dalitz pair invariant mass of virtual photon Another Idea Start from Dalitz decay Calculate inv. mass distribution of Dalitz pairs‘ Now direct photons Any source of real  produces virtual  with very low mass Rate and mass distribution given by same formula –No phase space factor for m ee << p T photon 00   00  e+e+ e-e- 

12. Torsten Dahms - Stony Brook University12 In practice 0-30 Material conversion pairs removed by analysis cut Combinatorial background removed by mixed events Calculate ratios of various m ee bins to lowest one: R data If no direct photons: ratios correspond to Dalitz decays If excess: direct photons Fit of virtual photon shape to data in principle also possible (done for d+Au) From conventional measurement ÷ 200-300 MeV ÷ 140-200 R data ÷ 90-140

13. Torsten Dahms - Stony Brook University13 Comparison Agreement of all three methods within their errors Also internal conversions favor an excess above decay photons

14. Torsten Dahms - Stony Brook University14 The spectrum Compare to NLO pQCD –L.E. Gordon and W. Vogelsang –Phys. Rev. D48, 3136 (1993) Above (questionable) pQCD Compare to thermal model –D. d’Enterria, D. Peressounko –nucl-th/0503054 Data above thermal at high p T Data consistent with thermal+pQCD Needs confirmation from p+p measurement 2+1 hydro T 0 ave =360 MeV (T 0 max =570 MeV)  0 =0.15 fm/c

15. Torsten Dahms - Stony Brook University15 Internal Conversions: d+Au Internal conversion method provides smaller systematic errors But not as small as in the case of Au+Au collisions in Run-4 (Large background of external conversion on MVD detector in Run-3)

16. Torsten Dahms - Stony Brook University16 Internal Conversion: d+Au Internal conversion extends range of significant points to pt > 2 GeV Data agrees with pQCD predictions in full pt range  No indication for nuclear effects

17. Torsten Dahms - Stony Brook University17 Comparison: d+Au & Au+Au Direct photon spectrum measured in d+Au collisions and scaled with agrees pretty well with spectrum measured in Au+Au at high p T There is room for thermal photons Systematic errors are still too big to decide whether excess seen in Au+Au can be assign to thermal source

18. Torsten Dahms - Stony Brook University18 Summary Photons are penetrating hard and soft probes for relativistic heavy ion collisions Conventional Calorimeter measurement –Systematic uncertainties at low p T too large to make definite statement about thermal photon contribution New methods show excess above decay photons Consistent with each other Internal conversions –Promising new technique to measure direct photons –Thermal photon scenario consistent for pT<3GeV/c –Because of the large systematic errors comparison of binary scaled d+Au spectrum with Au+Au does not allow to make a statement on the origin of the excess above pQCD observed in Au+Au –Same analysis of p+p is needed as confirmation

19. Torsten Dahms - Stony Brook University19

20. Backup

21. Torsten Dahms - Stony Brook University21 The PHENIX experiment Beam Pipe West ArmEast Arm γ e+e+ e-e- e+e+ e-e- γ Collision Vertex e+e+ e-e- γ electrons: momentum reconstruction (1% resolution) particle ID: RICH (loose cuts because clean signature of conversion peak) same or opposite arms: different pT acceptance photons: EMCal (loose cuts  high efficiency ~ 98%) track reconstruction assumes vertex in the interaction point  conversion at radius r≠0: e+e- pairs ‘acquire’ an opening angle  they acquire an invariant mass m =  B dl ~ r > 0 if r=4 cm (beam pipe) m =20 MeV

22. Torsten Dahms - Stony Brook University22 Invariant e + e - mass spectrum of Run 4 Au+Au: Dalitz decays beampipe conversions air conversions & combinatorial background Conversion pairs are created off-vertex Track reconstruction produces apparent opening angle Leads to apparent mass ~20MeV/c 2

23. Torsten Dahms - Stony Brook University23 Dalitz decays have a real opening angle due to the π 0 mass Conversion pairs have small intrinsic opening angle –magnetic field produces opening of the pair in azimuth direction –orientation perpendicular to the magnetic field Pair properties z y x e+e+ e-e- B Conversion pair z y x e+e+ e-e- B Dalitz decay MC Simulation all pairs dalitz decay beam pipe conversions MC Simulation all pairs dalitz decay beam pipe conversions

24. Torsten Dahms - Stony Brook University24 Simulations: N γ hadr (p T ) and N γ π 0 tag (p T ) Inclusive photon spectrum –π 0, η → γe + e - π 0 parameterization from measured data η from m T scaling, yield normalized at high p T (0.45 from measurement) –Use Dalitz decay (π 0 → γ γ ~ π 0 → γ γ* → γe + e - for p T > 0.8 GeV/c) All e + e - (from π 0, η) in the acceptance  p T spectrum of e + e - If γ from π 0 is also in acceptance  p T spectrum of e + e - from π 0 all e + e - pairs e + e - pairs from π 0 1/N evt dN/dp T [c/GeV]

25. Torsten Dahms - Stony Brook University25 Cocktail ingredients (pp):  0 most important: get the  0 right (>80 %), assumption:  0 = (  + +  - )/2 parameterize PHENIX pion data: most relevant: the  meson (Dalitz & conversion) also considered:  ’  use mT scaling for the spectral shape, i.e. normalization from meson/  0 at high pT as measured (e.g.  0 = 0.45±0.10)

26. Torsten Dahms - Stony Brook University26 Gale QM05