1. Ralf Averbeck Department of Physics & Astronomy High Energy Dilepton Experiments Experiments @ SPS

2. Ralf Averbeck, 2 Lepton-pair physics: topics chiral symmetry restoration continuum enhancement modification of vector mesons thermal radiation suppression (enhancement) l known sources of lepton pairs l modifications expected due to the QCD phase transition(s) l lepton pairs are –rich in physics –experimentally challenging l emitted over the full evolution of the collision l reach detectors undistorted from strong FSI

3. Ralf Averbeck, 3 SPS @ CERN l SuperProtonSynchrotron (since 1976) l parameters –circumference: 6.9 km –beams for fixed target experiments –protons up to 450 GeV/c –lead up to 158 GeV/c l past –SppS proton-antiproton collider  discovery of vector bosons W ±, Z l future –injector for LHC l experiments –Switzerland: west area (WA) –France: north area (NA)  dileptons speak french!

4. Ralf Averbeck, 4 Dilepton experiments @ SPS ExperimentSystemMass rangePublications HELIOS-1  ee p-Be (86)low massZ.Phys. C68 (1995) 64 HELIOS-3  p-W,S-W (92)low & lntermediate E.Phys.J. C13(2000)433 CERESeepBe, pAu, SAu (92/93) Pb-Au (95) Pb-Au (96) low massPRL (1995) 1272 Phys.Lett. B (1998) 405 Nucl.Phys. A661 (1999) 23 CERES-2eePb-Au 40 GeV (99) Pb-Au 158 GeV (2000) low massPRL 91 (2002) 42301 preliminary data 2004 NA38/ NA50  p-A, S-Cu, S-U, Pb-Pblow (high m T ) intermediate E.Phys.J. C13 (2000) 69 E.Phys.J. C14 (2000) 443 NA60  p-A, In-In (2002,2003) p-A (2004) >2m  PRL 96 (2006) 162302

5. Ralf Averbeck, 5 The CERES/NA45 experiment

6. Ralf Averbeck, 6 Experimental setup: CERES-1

7. Ralf Averbeck, 7 Target region 13 l segmented target l 13 Au disks (thickness: 25  m; diameter: 600  m) l Silicon drift chambers: l provide vertex:  z = 216  m l provide event multiplicity (  = 1.0 – 3.9) l powerful tool to recognize conversions at the target

8. Ralf Averbeck, 8 Electron identification: RICH l main tool for electron ID l use the number of hits per ring (and their analog sum) to recognize single and double rings

9. Ralf Averbeck, 9 Dielectron analysis strategy

10. Ralf Averbeck, 10 l dielectron mass spectra and expectation from a ‘cocktail’ of known sources Dalitz decays of neutral mesons (   →  e + e - and  ’  l dielectron decays of vector mesons (  → e + e - ) l semileptonic decays of particles carrying charm quarks  dielectron production in ‘pp’ and ‘pA’ collisions at SPS well understood in terms of known hadronic sources! e + e - in p+Be & p+Au collisions

11. Ralf Averbeck, 11 What about heavy-ion collisions? CERES PRL 92 (95) 1272 l discovery of low mass e + e - enhancement in 1995 l significant excess in S-Au (factor ~5 for m>200 MeV)

12. Ralf Averbeck, 12 As heavy as it gets: Pb+Au l dielectron excess at low and intermediate masses in HI collisions is well established onset at ~2 m    -  annihilation? maximum below  meson near 400 MeV  hint for modified  meson in dense matter     e-e- e+e+ CERES Eur.Phys.Jour. C41(2005)475

13. Ralf Averbeck, 13  -  annihilation: theoretical approaches low mass enhancement due to  annihilation? spectral shape dominated  meson vacuum  l vacuum values of width and mass in-medium  l Brown-Rho scaling –dropping masses as chiral symmetry is restored l Rapp-Wambach melting resonances –collision broadening of spectral function –only indirectly related to CSR l medium modifications driven by baryon density l model space-time evolution of collision     e-e- e+e+

14. Ralf Averbeck, 14 Theory versus CERES-1 data l attempt to attribute the observed excess to vacuum  meson ( ) –inconsistent with data –overshoot in  region –undershoots @ low mass modification  meson –needed to describe data –data do not distinguish between –broadening or melting of  - meson (Rapp-Wambach) –dropping masses (Brown-Rho) l indication for medium modifications, but data are not accurate enough to distinguish models largest discrimination between  and   need mass resolution!

15. Ralf Averbeck, 15 CERES-1  CERES-2 l addition of a TPC to CERES l improved momentum resolution l improved mass resolution l dE/dx  hadron identification and improved electron ID l inhomogeneous magnetic field  a nightmare to calibrate!

16. Ralf Averbeck, 16 CERES-2 result l the CERES-1 results persists l strong enhancement in the low-mass region l enhancement factor (0.2 <m < 1.1 GeV/c 2 )  3.1 ± 0.3 (stat.) l but the improvement in mass resolution isn ’ t ‘ outrageous ’

17. Ralf Averbeck, 17 Dropping mass, broadening, or thermal radiation  dropping  meson mass (Brown et al) * in-medium modifications of  :  broadening  spectral shape (Rapp and Wambach)  thermal radiation (e + e - yield calculated from qbarq ann. In pQCD B.Kämpfer et al) l interpretations invoke l  +  -     *  e + e - l thermal radiation from hadron gas l vacuum  not enough to reproduce the data

18. Ralf Averbeck, 18 PRL 91 (2003) 042301 CERES @ low energy (40 GeV/c) l data taking in 1999 and 2000 l improved mass resolution l improved background rejection l results remain statistics limited l Pb-Au at 40 AGeV l enhancement for m ee > 0.2 GeV/c 2 –5.9±1.5(stat)±1.2(sys)±1.8(decay) strong enhancement at lower  s or larger baryon density vacuum  Brown-Rho scaling broadening of 

19. Ralf Averbeck, 19 And what about p T dependence? l low mass e + e - enhancement at low p T qualitatively in a agreement with  annihilation l p T distribution has little discriminative power m ee <0.2 GeV/c 2 0.2<m ee <0.7 GeV/c 2 m ee >0.7 GeV/c 2 hadron cocktail Brown-Rho scaling broadening of 

20. Ralf Averbeck, 20 Centrality dependence of excess l naïve expectation: quadratic multiplicity dependence l medium radiation  particle density squared l more realistic: smaller than quadratic increase l density profile in transverse plane l life time of reaction volume F=yield/cocktail m ee <0.2 GeV/c 2 0.2<m ee <0.6GeV/c 2 m ee >0.6 GeV/c 2 CERES p T > 200 MeV/c 1995/96 2000  N ch strong centrality dependence  challenge for theory !

21. Ralf Averbeck, 21 What did we get from CERES? l first systematic study of e + e - production in elementary and HI collisions at SPS energies l pp and pA collisions are consistent with the expectation from known hadronic sources l a strong low-mass low-p T enhancement is observed in HI collisions  consistent with in-medium modification of the  meson  data can’t distinguish between two scenarios  dropping  mass as direct consequence of CSR  collisional broadening of  in dense medium l WHAT IS NEEDED FOR PROGRESS? l STATISTICS l MASS RESOLUTION

22. Ralf Averbeck, 22 How to overcome these limitations l more statistics l run forever  not an option l higher interaction rate –higher beam intensity –thicker target l needed to tolerate this –extremely selective hardware trigger –reduced sensitivity to secondary interactions, e.g. in target l  can’t be done with dielectrons as a probe, but dimuons are just fine! l better mass resolution l stronger magnetic field l detectors with better position resolution l  silicon tracker embedded in strong magnetic field!

23. Ralf Averbeck, 23 The NA60 experiment l a huge hadron absorber and muon spectrometer (and trigger!) l and a tiny, high resolution, radiation hard vertex spectrometer

24. Ralf Averbeck, 24 Standard  +  - detection: NA50 thick hadron absorber to reject hadronic background l trigger system based on fast detectors to select muon candidates (1 in 10 4 PbPb collisions at SPS energy) l muon tracks reconstructed by a spectrometer (tracking detectors+magnetic field) l extrapolate muon tracks back to the target taking into account multiple scattering and energy loss, but … poor reconstruction of interaction vertex (  z ~10 cm) poor mass resolution (80 MeV at the  ) Muon Other hadron absorber muon trigger and tracking target beam magnetic field

25. Ralf Averbeck, 25 2.5 T dipole magnet hadron absorber targets beam tracker vertex tracker muon trigger and tracking magnetic field Muon Other A step forward: the NA60 case or ! matching of muon tracks l origin of muons can be determined accurately l improved dimuon mass resolution

26. Ralf Averbeck, 26 DIPOLE MAGNET 2.5 T HADRON ABSORBER TARGETS ~40 cm 1 cm The NA60 pixel vertex spectrometer l 12 tracking points with good acceptance l 8 small 4-chip planes l 8 large 8-chip planes in 4 tracking stations l ~3% X 0 per plane 750  m Si readout chip 300  m Si sensor l ceramic hybrid l 800000 readout channels in 96 pixel assemblies

27. Ralf Averbeck, 27 Beam Tracker sensors windows  z ~ 200  m along the beam direction Good vertex identification with  4 tracks X Y Extremely clean target identification (Log scale!) Vertexing in NA60 Resolution ~ 10 - 20  m in the transverse plane

28. Ralf Averbeck, 28 Contributions to mass resolution l two components l multiple scattering in the hadron absorber –dominant at low momentum l tracking accuracy –dominant at high momentum l high mass dimuons (~3 GeV/c 2 ) l absorber doesn’t matter l low mass dimuons (~1 GeV/c 2 ) l absorber is crucial l momentum measurement before the absorber promises huge improvement in mass resolution l  track matching is critical for high resolution low mass dimuon measurements!

29. Ralf Averbeck, 29 Muon track matching l track matching has to be done in l position space l momentum space l to be most effective l  the pixel telescope has to be a spectrometer! Muon spectrometerPixel telescopeAbsorber Measured points

30. Ralf Averbeck, 30 6500 A 4000 A dN/dM  (Events/50 MeV) (80% of collected statistics) (100% of collected statistics) Vertex selection and muon track matching  M (  )  80 MeV  M (J/  )  100 MeV  M (  )  20 MeV  (1020)  (1020)   M (J/  )  70 MeV Improvement in mass resolution l unlike sign dimuon mass distribution before quality cuts and without muon track matching l drastic improvement in mass resolution l still a large unphysical background

31. Ralf Averbeck, 31 hadron absorber muon trigger and tracking target fake correct Hadron absorber Muon spectrometer Nothing is perfect: fake matches fake match:  matched to wrong track in pixel telescope l important in high multiplicity events l how to deal with fake matches keep track with best  2 (but is is right?) l embedding of muon tracks into other event l identify fake matches and determine the fraction of these relative to correct matches as function of –centrality –transverse momentum

32. Ralf Averbeck, 32 Event mixing: like-sign pairs l compare measured and mixed like-sign pairs l accuracy in NA60: ~1% over the full mass range

33. Ralf Averbeck, 33    Final mass spectra (m<2 GeV/c 2 ) WOW!

34. Ralf Averbeck, 34 The low-mass region BR = 5.8x10 -6 ! l enormous statistics! l fantastic resolution!

35. Ralf Averbeck, 35 l ω and  : fix yields such as to get, after subtraction, a smooth underlying continuum l  : (  ) set upper limit, defined by “ saturating ” the measured yield in the mass region close to 0.2 GeV (lower limit for excess). (  ) use yield measured for p T > 1.4 GeV/c Cocktail subtraction (without  ) l how to nail down an unknown source? l  try to find excess above cocktail without fit constraints

36. Ralf Averbeck, 36 Clear excess above the cocktail , centered at the nominal  pole and rising with centrality Excess even more pronounced at low p T No cocktail  and no DD subtracted data – cocktail (all p T ) Excess versus centrality

37. Ralf Averbeck, 37 Quantify the peak and the broad symmetric continuum with a mass interval C around the peak (0.64 <M<0.84 GeV) and two equal side bins L, U continuum = 3/2(L+U) peak = C-1/2(L+U) Peak/cocktail  drops by a factor  2 from peripheral to central: the peak seen is not the cocktail  nontrivial changes of all three variables at dN ch /dy>100 ? peak/  continuum/  peak/continuum Fine analysis in 12 centrality bins Excess shape versus centrality

38. Ralf Averbeck, 38 data consistent with broadening of  ( RW), mass shift (BR) not needed Comparison with prominent models l Rapp & Wambach l hadronic model with strong broadening but no mass shift l Brown & Rho l dropping mass due to dropping chiral condensate calculations for all scenarios in In-In for dN ch /d  = 140 (Rapp et al.) l spectral functions after acceptance filtering, averaged over space-time and momenta l yields normalized to data for m < 0.9 GeV

39. Ralf Averbeck, 39 Role of baryons l improved model calculation (Rapp & van Hees) l fireball dynamics 4  processes l absolute normalization! towards high p T the vacuum  becomes more important (Rapp/van Hees; Renk/Ruppert) l without baryons –not enough broadening –lack of strength below the  peak

40. Ralf Averbeck, 40 The high mass region (M>1GeV) l hadron-parton duality Rapp / van Hees Ruppert / Renk l dominant at high M l hadronic processes 4  l dominant at high M l partonic processes l mainly qqbar annihilation

41. Ralf Averbeck, 41 central collisions M (GeV/c 2 ) Intermediate mass region (IMR) l NA50: excess observed in IMR in central Pb-Pb collisions l charm enhancement? l thermal radiation? l answering this question was one of the main motivations for building NA60

42. Ralf Averbeck, 42 D0D0 K-K- ++ e D0D0 100 m Disentangling the sources in the IMR l charm quark-antiquark pairs are mainly produced in hard scattering processes in the earliest phase of the collisions K+K+ -- charmed hadrons are “long” lived  identify the typical offset (“displaced vertex”) of D-meson decays (~100  m) need superb vertexing accuracy (20-30  m in the transverse plane)  NA60

43. Ralf Averbeck, 43 How well does this work? l measure for vertex displacement l primary vertex resolution l momentum dependence of secondary vertex resolutions l  “dimuon weighted offset” l charm decays (D mesons)  displaced J/    prompt l vertex tracking is well under control!

44. Ralf Averbeck, 44 IMR excess: enhanced charm? l approach l fix the prompt contribution to the expected Drell-Yan yield l check whether the offset distribution is consistent with charm dN/dΔ New alignment l charm can’t describe the small offset region!

45. Ralf Averbeck, 45 How many prompt pairs are needed? l approach l fit offset distribution with both charm and prompt contributions as free parameters l prompt component l ~2.4 times larger than Drell-Yan contribution l charm component l ~70% of the yield extrapolated from NA50’s p-A data 4000 A,  2 < 3 6500 A,  2 < 3

46. Ralf Averbeck, 46 Decomposition of mass spectrum IMR: 1.16 < M < 2.56 GeV/c 2 (between  and J/  ) l definition of excess l excess = signal – [ Drell-Yan (1.0 ± 0.1) + Charm (0.7±0.15) ]

47. Ralf Averbeck, 47 Centrality & p T dependence of IMR excess l increase more than proportional to N part l but also more than proportional to N coll ! Excess/N participants (arb. scale) 4000kA + 6500 kA data, corrected for acceptance l p T distribution is significantly softer than the (hard) Drell-Yan contribution

48. Ralf Averbeck, 48 fit in 0.5<p T <2 GeV/c More detailed look at p T dependence l investigate excess in different mass regions as function of m T l fit with exponential function (shown for IMR) l extract T eff slope parameter l ~ 190 MeV l is this related to temperature? l if so, this is close to the critical temperature at which the QCD phase transition occurs

49. Ralf Averbeck, 49 Interpretation of T eff l interpretation of T eff from fitting to exp(-m T /T eff ) l static source: T eff interpreted as the source temperature l radially expanding source: –T eff reflects temperature and flow velocity –T eff dependens on the m T range –large p T limit: common to all hadrons –low p T limit: mass ordering of hadrons l final spectra: space-time history T i →T fo & emission time l hadrons –interact strongly –freeze out at different times depending on cross section with pions –T eff  temperature and flow velocity at thermal freeze out l dileptons –do not interact strongly –decouple from medium after emission –T eff  temperature and velocity evolution averaged over emission time

50. Ralf Averbeck, 50 158 AGeV Central collisions Pb-Pb In-In Si-Si C-C pp Mass ordering of hadronic slopes l separation of thermal and collective motion l reminder l blast wave fit to all hadrons simultaneously l simplest approach l slope of vs. m is related to radial expansion l baseline is related to thermal motion l works (at least qualitatively) at SPS

51. Ralf Averbeck, 51 v T = 0.1 v T = 0.2 v T = 0.3 v T = 0.4 v T = 0.5 (specific for In-In: Dusling et al.) Example of hydrodynamic evolution hadron phase parton phase l monotonic decrease of T from l early times to late times l medium center to edge l monotonic increase of v T from l early times to late times l medium center to edge l dileptons may allow to disentangle emission times l early emission (parton phase) –large T, small v T l late emission (hadron phase) –small T, large v T

52. Ralf Averbeck, 52 NA60 analysis of m T spectra in In-In l decomposition of low mass region contributions of mesons ( , ,  ) continuum plus  meson extraction of vacuum  hadron m T spectra for l , ,  vacuum  l dilepton m T spectra for l low mass excess l intermediate mass excess Phys. Rev. Lett. 96 (2006) 162302

53. Ralf Averbeck, 53 Examples of m T distributions l variation with mass are obvious

54. Ralf Averbeck, 54 Dilepton T eff systematics hadrons (  ) l T eff depends on mass T eff smaller for , decouples early T eff large for , decouples late l low mass excess l clear flow effect visible l follows trend set by hadrons l possible late emission l intermediate mass excess l no mass dependence l indication for early emission

55. Ralf Averbeck, 55 What did we get from NA60? l high statistics & high precision dimuon spectra l decomposition of mass spectra into “sources” gives access to in-medium  spectral function data consistent with broadening of the  data do not require mass shift of the  l large prompt component at intermediate masses l dimuon m T spectra promise to separate time scales l low mass dimuons shows clear flow contribution indicating late emission l intermediate mass dimuons show no flow contribution hinting toward early emission