1. Dileptons @RHIC a clock and a thermometer of relativistic heavy ion collisions Alberica Toia for the PHENIX Collaboration Stony Brook University / CERN International School of Nuclear Physics “Heavy ion collisions from the Coulomb barrier to the Quark-Gluon Plasma” Erice, September 16-24 2008

2. Dileptons: radiation from the medium (KEK E235)CERESDLS NA60HADESCBMPHENIX 10158[A GeV] 17[GeV]√s NN 200 // ALICE [A TeV] time hard parton scattering Au hadronization freeze-out formation and thermalization of quark-gluon matter? Space Time expansion Jet cc  e   p K    direct probes large emission rate in hot and dense matter light vector mesons: signals of chiral transition? Measurement at RHIC: different energy different initial conditions different hydro evolution hard processes have larger cross sections collider geometry

3. Dilepton Signal Extraction arXiv: 0706.3034 Au+Au Single electron –Track reconstruction –Electron Identification (RICH + EMCal) Background sources – Combinatorial background – Material conversion pairs – Additional correlated background Visible in p+p collisions Cross pairs from decays with 4 electrons in the final state Pairs in same jet or back-to- back jet Real signal – di-electron continuum p+p arXiv: 0802.0050

4. Hadronic Cocktail Calculation arXiv: 0802.0050 Remaining pairs after background subtraction – Real signal + Hadron decay components Parameterized PHENIX  0 data with assumption of  0 = (  + +  - )/2 Other mesons are fit with m T scaling of π 0 parameterization p T → √(p T 2 +m meson 2 - m π 2 ) fit the normalization constant  All mesons m T scale!!! Hadronic cocktail was well tuned to individually measured yield of mesons in PHENIX for both p+p and Au+Au collisions. Mass distributions from hadron decays are simulated by Monte Carlo. –  0, ,  ’, , , , J/ ,  ’ Effects on real data are implemented. – PHENIX acceptance, detector effect, efficiencies …

5. Cocktail Comparison arXiv: 0802.0050arXiv: 0706.3034 Au+Aup+p – Excellent agreement with cocktail Au+Au – Large enhancement in low mass region – Integrated yield in 150MeV < m ee < 750MeV data/cocktail = 3.4 ± 0.2(stat) ± 1.3(sys) ± 0.7(model)

6. Intermediate Mass: yield increase proportional to N coll  charm follows binary scaling Centrality Dependency submitted to Phys. Rev. Lett arXiv:0706.3034 LOW MASS INTERMEDIATE MASS  0 region: Agreement with cocktail Low Mass: yield increases faster than proportional to N part  enhancement from binary annihilation (ππ or qq) ?

7. Source of the excess? Freeze-out Cocktail + “random” charm +  spectral function Low mass M>0.4GeV/c 2 : some calculations OK M<0.4GeV/c 2 : data still above theory calculations Intermediate mass Random charm + thermal partonic may work

8. 0 < p T < 8 GeV/c0 < p T < 0.7 GeV/c 0.7 < p T < 1.5 GeV/c1.5 < p T < 8 GeV/c PHENIX Preliminary p T -Sliced Mass Spectra ○ Au+Au ● p+p Normalized by the yield in m ee < 100MeV The low mass enhancement decreases with higher p T Different shape at low and high p T  low mass enhancement does not contribute to m 1 GeV/c?

9. p+pAu+Au (MB) 1 < p T < 2 GeV/c 2 < p T < 3 GeV/c 3 < p T < 4 GeV/c 4 < p T < 5 GeV/c p+p –Good agreement between real and cocktail –Small excess at higher p T Au+Au – Good agreement in M ee < 50MeV/c 2 – Enhancement is clearly seen above 100MeV/c 2. Low mass & High p T region

10. Sources of e+e- pairs Dalitz:  0   e+e-    e+e-    0 e+e-    e+e- Direct:   e+e-   e+e-   e+e- J/   e+e-  ’  e+e- Heavy flavor: cc  e+e- +X bb  e+e- +X Drell-Yan: qq  e+e- cocktail Gluon Compton q  g q e+e+ e-e- Rising at low mass (~1/m): due to Dalitz decays of baryonic/mesonic resonances a1   a1     e+e- N*  N  N*  N  *  Ne+e- where the underlying process is: Quarks and gluons can be “free” (in QGP) or “hidden” in hadrons Theoretical predictions:     e+e- threshold at 2m  ~300MeV qqbar    e+e- “threshold-like” m q < 150 MeV? Much smaller than cocktail Cocktail Hadron Gas QGP (qq   *  e+e-) Mass (GeV/c 2 )

11. Direct Photons Measurement Any source of real  can emit  * with very low mass. If the Q 2 (=m 2 ) of virtual photon is sufficiently small, the source strength should be the same The ratio of real photon and quasi-real photon can be calculated by QED  Real photon yield can be measured from virtual photon yield, which is observed as low mass e + e - pairs Kroll-Wada formula S : Process dependent factor Case of Hadrons Obviously S = 0 at M ee > M hadron Case of  * If p T 2 >>M ee 2 Possible to separate hadron decay components from real signal in the proper mass window. q  g q e+e+ e-e-

12. Determination of  * fraction, r r : direct  * /inclusive  * Direct  * /inclusive  * is determined by fitting the following function for each p T bin. Reminder : f direct is given by Kroll-Wada formula with S = 1. Fit in 80-300MeV/c 2 gives – Assuming direct  * mass shape  2 /NDF=11.6/10 – Assuming  shape instead of direct  * shape  2 /NDF=21.1/10 Twice as much as measured  yield Assumption of direct  * is favorable.

13. direct  * /inclusive  * μ = 0.5p T μ = 1.0p T μ = 2.0p T Base line Curves : NLO pQCD calculations with different theoretical scales done by W. Vogelsang. p+p Consistent with NLO pQCD – better agreement with small µ Au+Au Clear enhancement above NLO pQCD p+pAu+Au

14. Direct Photon Spectra The virtual direct photon fraction is converted to the direct photon yield. p+p – First measurement in 1-4GeV/c – Consistent with NLO pQCD and with EmCal method – Serves as a crucial reference Au+Au – Above binary scaled NLO pQCD – Excess comes from thermal photons? NLO pQCD (W. Vogelsang) Fit to pp exp + TAA scaled pp scaled ppexponential

15. 1 st measurement of Thermal Radiation Au+Au = pQCD + exp.  T = 221  23 (stat)  18 (sys) Initial temperatures and times from theoretical model fits to data: –0.15 fm/c, 590 MeV (d ’ Enterria et al.) –0.17 fm/c, 580 MeV (Rasanen et al.) –0.2 fm/c, 450-660 MeV (Srivastava et al.) –0.33 fm/c, 370 MeV (Turbide et al.) –0.6 fm/c, 370 MeV (Liu et al.) –0.5 fm/c, 300 MeV (Alam et al.) From data: T ini > 220 MeV > T C From models: T ini = 300 to 600 MeV   = 0.15 to 0.5 fm/c D.d’Enterria, D.Peressounko, Eur.Phys.J.C 46 (2006)

16. Dilepton Spectra p+p Au+Au p+p –Agreement with cocktail Au+Au –p T >1GeV/c: small excess  internal conversion of direct photons –p T <1GeV/c: large excess  other source???

17. Summary p+p LOW MASS: Excellent agreement with hadronic decay cocktail INTERMEDIATE MASS: Extract charm and bottom cross sections σ cc = 544 ± 39 (stat) ± 142 (syst) ± 200 (model) μb σ bb = 3.9 ± 2.4 (stat) +3/-2 (syst) μb DIRECT PHOTONS p+p in agreement with pQCD Au+Au LOW MASS: Enhancement above the cocktail expectations: 3.4±0.2(stat.) ±1.3(syst.)±0.7(model) Centrality dependency: increase faster than N part Enhancement concentrated at low p T INTERMEDIATE MASS: Coincidence agreement with PYTHIA Room for thermal radiation? DIRECT PHOTONS: Dielectron mass shape for p T > 1 GeV and m ee < 300MeV consistent with internal conversions of virtual photons Au+Au above pQCD excess with inv. slope: T eff = 221 ± 23 (stat) ± 18 (syst) well described by hydrodynamical models with initial coonditions of T init =300–600 MeV at τ 0 = 0.15–0.6 fm/c First dielectron continuum measurement at RHIC –Despite of low signal/BG –Thanks to high statistics –Very good understanding of background normalization HBD upgrade will reduce background  great improvement of systematic and statistical uncertainty (LMR) Silicon Vertex detector will distinguish charm from prompt contribution (IMR) 17

18. Backup

19. The matter is so dense that it melts(?) J/  (and regenerates it ?) sQGP @ RHIC The matter is so opaque that even a 20 GeV  0 is stopped The matter is so strongly coupled that even heavy quarks flow The matter is so dense that it modifies the shape of jets PHENIX preliminary strongly interacting Quark-Gluon Plasma (sQGP) in HI collisions at RHIC The matter is so dense that even heavy quarks are stopped What does it emit? What is the temperature?

20. Not a new idea… The idea of measuring direct photon via low mass lepton pair is not new one. It is as old as the concept of direct photon. This method is first tried at CERN ISR in search for direct photon in p+p at 55GeV. They look for e+e- pairs for 200<m<500 MeV, and they set one of the most stringent limit on direct photon production at low pT Later, UA1 measured low mass muon pairs and deduced the direct photon cross section. J.H.Cobb et al., PL 78B, 519 (1978) Credit to try it in PHENIX goes to Y.Akiba

21. Previous measurements The enhancement is concentrated at low p T CERES measured an excess of dielectron pairs, confirmed by NA60, rising faster than linear with centrality attributed to in- medium modification of the  spectral function from  annihilation. CERES NA60

22. Understanding the p T dependency Comparison with cocktail Single exponential fit: –Low-p T : 0<m T <1 GeV –High-p T : 1<m T <2 GeV 2-components fits –2exponentials –m T -scaling of  0 + exponential Low p T : –inverse slope of ~ 120MeV –accounts for most of the yield

23. Extract 2 components 2 EXPONENTIALSHAGEDORN + EXPONENTIAL We fit the sum of 2 exponentials (a*exponential1 + b*exponential2) We fit Hagedorn to M ee <100MeV (  0 -dominated) Then we fit (a*m T -scaling + exponential) to the other mass bins Because of their different curvature, m T -scaling and the exponential account for more or less of the yield in the low-p T region.

24. Yields and Slopes Intermediate p T : inverse slope increase with mass, consistent with radial flow Low p T : inverse slope of ~ 120MeV accounts for most of the yield SLOPESYIELDS Total yield (DATA) 2expo fit m T -scaling +expo fit Low-p T yield

25. Theory Comparison II Freeze-out Cocktail + “random” charm +  spectral function Low mass M>0.4GeV/c 2 : some calculations OK M<0.4GeV/c 2 : not reproduced Intermediate mass Random charm + thermal partonic may work Low-p T slope not reproduced Gluon Compton q  g q e+e+ e-e- q q HADRONIC PARTONIC  -  annihilation q-q annihilation R.Rapp + H.vanHees K.Dusling + I.Zahed E.Bratkovskaja + W.Cassing

26. Theory Comparison II Calculations from R.Rapp & H.vanHees K.Dusling & I.Zahed E.Bratovskaja & W.Cassing (in 4  )

27. Questions 1. Enhancement at M<2M  If pions are massless can  annihilation produce ee with M<300MeV? 2. Enhancement at low p T, with T~120 MeV and now flow Is the same low-p T enhancement seen at SPS never reproduced by theory? Different initial temperature Different system evolution Do we miss something in the system evolution which may have different relevance at SPS and at RHIC? RHIC SPS

28. PHENIX (Pioneering High Energy Nuclear Interaction eXperiment) 2 central arms: electrons, photons, hadrons –charmonium J/ ,  ’  e  e  –vector meson   e  e  –high p T       –direct photons –open charm –hadron physics 2 muon arms: muons –“onium” J/ ,  ’,      –vector meson      –open charm combined central and muon arms: charm production DD  e  global detectors forward energy and multiplicity –event characterization Au-Au & p-p spin PC1 PC3 DC magnetic field & tracking detectors e+e+ ee   designed to measure rare probes: + high rate capability & granularity + good mass resolution and particle ID - limited acceptance

29. Photon conversion rejection Conversion removed with orientation angle of the pair in the magnetic field Photon conversion   e+e- at r≠0 have m≠0 (artifact of PHENIX tracking: no tracking before the field) effect low mass region have to be removed Beampipe MVD support structures r ~ m ee Inclusive Removed by phiV cut After phiV cut z y x e+e+ e-e- B Conversion pair z y x e+e+ e-e- B Dalitz decay

30. Photon conversion cut No cut M<30 MeV &  V <0.25 & M<600 MeV &  V <0.04 M<600 MeV &  V <0.06 M<600 MeV &  V <0.08 M<600 MeV &  V <0.10 M<600 MeV &  V <0.12 M<600 MeV &  V <0.14 M<600 MeV &  V <0.20 M<600 MeV &  V <0.40

31. Physical background Semi-correlated Background π0π0 π0π0 e+e+ e-e- e+e+ e-e- γ γ π0π0 e-e- γ e+e+ X  0   *  e+e- e+e- “jets” z y x e+e+ e-e- B Conversion pair z y x e+e+ e-e- B Dalitz decay Photon conversion   e+e- at r≠0 have m≠0 (artifact of PHENIX tracking) Conversion removed with orientation angle of the pair in the magnetic field Background is charge-independent Calculate the shape with MC Normalize to the like-sign spectra  Good description of the data arXiv: 0802.0050

32. Combinatorial Background PHENIX 2 arm spectrometer acceptance: dN like /dm ≠ dN unlike /dm different shape  need event mixing (like/unlike differences preserved) Use Like sign as a cross check for the shape and to determine normalization Small signal in like sign at low mass  N++ and N–- estimated from the mixed events like sign B++ and B-- normalized at high mass (> 700 MeV) Normalization: 2√N++ N--  Uncertainty due to statistics of N++ and N--: 0.12% Correction for asymmetry of pair cut K=k+-/√k++ k-- = 1.004 Systematic error (conservative): 0.2% TOTAL SYSTEMATIC ERROR = 0.25% LIKE SIGN SPECTRA Use same event topology (centrality, vertex, reaction plane) Remove every unphysical correlation

33. Comparison of BG subtraction Methods 33 to determine syst. uncertainty: spread of two extreme cases (blue & orange): 5-10% Monte Carlo method Like sign method (with some variations) give consistent results over the full invariant mass range

34. Acceptance charge/pT 00 00 z vertex Define acceptance filter (from real data) Correct only for efficiency IN the acceptance “Correct” theory predictions IN the acceptance Single electron pT > 200 MeV  Pair mT > 400 MeV Not an analysis cut, but a constrain from the magnetic field mass pTpT

35. Cross check Converter Method We know precise radiation length (X 0 ) of each detector material The photonic electron yield can be measured by increase of additional material (photon converter was installed) The non-photonic electron yield does not increase Photonic single electron: x 2.3 Inclusive single electron :x 1.6 Combinatorial pairs :x 2.5 Photon Converter (Brass: 1.7% X 0 ) N e Electron yield Material amounts:  0 0.4%1.7% Dalitz : 0.8% X 0 equivalent radiation length 0 With converter W/O converter 0.8% Non-photonic Photonic converter