1. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 1 Probing deconfinement with quarkonia : new answers to old questions Carlos Lourenço, CERN Workshop on Hot and Dense Matter, Mumbai, Feb. 2008 Outline: What are the “cold nuclear matter effects” in charmonium absorption ? How do they affect the SPS J/  and  ’ suppression patterns ? [ work being done in collaboration with Ramona Vogt and Hermine K. Wöhri ] What gets you into trouble is not what you don’t know but what you think you know Mark Twain (Larry at QM08)

2. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 2 QGP ? We study the bulk QCD matter produced in HI collisions by seeing how it affects well understood probes as a function of the temperature of the system (centrality of the collisions) Calibrated “probe source” Matter under study Calibrated “probe meter” Calibrated heat source Probe “Seeing” the QCD matter formed in heavy-ion collisions

3. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 3 Challenge: find the good probes of QCD matter vacuum QGP hadronic matter The good QCD matter probes should be: Heavy quarkonia (J/ ,  ’, ,  ’, etc) are very good QCD matter probes ! Well understood in “pp collisions” Slightly affected by the hadronic matter, in a well understood way, which can be accounted for Strongly affected by the deconfined QCD medium...

4. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 4 reference process Drell-Yan dimuons are not affected by the dense medium they cross The yield of J/  mesons (per DY dimuon) is “slightly smaller” in p-Pb collisions than in p-Be collisions; and is strongly suppressed in central Pb-Pb collisions Interpretation: strongly bound c-cbar pairs are “dissolved” by the QCD medium created in central Pb-Pb collisions at SPS energies p-Be p-Pb central Pb-Pb reference data J/  normal nuclear absorption curve S-U Pb-Pb p-A NA38 / NA51 / NA50 J/  suppression: the NA38/50/51 picture

5. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 5 We must have a robust and well understood reference baseline, in A-A collisions, with respect to which we can clearly and unambiguously identify patterns specific to the high-density medium produced in high-energy nuclear collisions What should we really expect in the absence of a deconfined QCD medium but accounting for all the other “standard” aspects of nuclear collisions? This requires : → Understanding the basic properties of quarkonium production in pp and p-A → A robust model to turn the p-A patterns into reliable A-A expectations Quarkonium studies in proton-nucleus collisions: why?

6. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 6 Charmonium absorption in p-nucleus collisions The J/  and  ’ production cross sections scale less than linearly with the number of target nucleons (unlike high-mass Drell-Yan dimuons) ’’ J/  NA50 p-A 400 GeV p-Pb @ 400 GeV  J/  ~ 105 MeV The Glauber model describes the J/  and  ’ “normal nuclear absorption” with a single parameter: the absorption cross section

7. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 7 From a global fit to the 400 and 450 GeV p-A data (16 independent measurements), NA50 determined the following absorption cross sections (with GRV94LO PDFs):  abs (J/  = 4.5 ± 0.5 mb ;  abs (  ’) = 8.3 ± 0.9 mb from production cross sections  abs (J/  = 4.2 ± 0.5 mb ;  abs (  ’) = 7.7 ± 0.9 mb from cross-section ratios (  /DY)  2 /ndf = 0.7  2 /ndf = 1.4 These calculations assume that the reduction of the production cross section per target nucleon is exclusively due to charmonium final-state absorption  abs = 4.5 ± 0.5 mb  abs = 8.3 ± 0.9 mb

8. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 8 It seems that the J/  absorption, at mid-rapidity, becomes weaker with increasing collision energy, at least between SPS and RHIC energies The 158 GeV p-A data of NA60 will clarify if the trend continues to lower energies J/  0 mb 3 mb Low x 2 ~ 0.003 (shadowing region) PHENIX J/  absorption in p-A collisions vs. collision energy Slide shown at HP06 Without nuclear effects on the PDFsWith nuclear effects on the PDFs

9. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 9 J/  The increase of  with p T is identical at 400, 800 and 920 GeV (at mid-rapidity)  Maybe the increase of  from NA50 to E866 to HERA-B to PHENIX is due to the increase of the average p T of the J/  when  s increases... And  strongly decreases at high x F where the J/  and  ’ have similar absorptions J/  absorption in p-A collisions vs. p T and x F NA50 E866 Slide shown at HP06

10. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 10 a R. Vogt, PRC 61 (2000) 035203, NP A700 (2002) 539 - R. Vogt, PRC 61 (2000) 035203, NP A700 (2002) 539 - K.G. Boreskov & A.B. Kaidalov, JETPL 77 (2003) 599 Models (with variants): 1.0 0.9 0.8 xFxF -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Given enough models… at least one should describe the data xFxF E86638.8 GeVBe/Fe/W E78938.8 GeVBe/C/Cu/W E77238.8 GeVH 2 /C/Ca/Fe/W NA5029.1 GeVBe/Al/Cu/Ag/W NA322.9 GeVH 2 /Pt E86638.8 GeVBe/Fe/W E78938.8 GeVBe/C/Cu/W E77238.8 GeVH 2 /C/Ca/Fe/W NA5029.1 GeVBe/Al/Cu/Ag/W NA322.9 GeVH 2 /Pt B&KB&K -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 1.0 0.9 0.8 0.7 HERA-B preliminary Vogt: final state absorption Slide shown at HP06

11. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 11 Nuclear effects on the PDFs The EKS 98 model gives significant anti-shadowing for charm production at the SPS Similar for p-Pb at 450 GeV and Pb-Pb at 158 GeV… but ~6% more J/  mesons are produced, per nucleon, in Pb-Pb than in p-Pb, if both are taken at 158 GeV

12. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 12 Initial state nuclear effects vs. final state absorption At SPS energies, the gluon anti-shadowing makes the J/  production cross section per nucleon increase from pp to p-Pb, if we would ignore final state absorption  abs = 0 mb  abs = 4 mb  abs = 7 mb EKS98 0<y*<1 ppp-Pb

13. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 13 p-Pb J/  EKS98 FGSo FGSh FGSl nDS nDSg DeFlorian and Sassot Eskola, Kolhinen and Salgado Frankfurt, Guzey and Strikman EKS98 is not the only available model of nuclear effects on the parton densities DeFlorian and Sassot predict no anti-shadowing for J/  production at SPS energies while the FGSo parameterization predicts stronger anti-shadowing than EKS98… The new EPS08 model gives more anti-shadowing at the SPS than EKS98…

14. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 14 Absorption of  ’ and  c states Approximate radii of the J/ ,  ’ and  c states: r(J/  ) = 0.25 fm; r(  ’) = 2 × r(J/  ); r(  c ) = 1.5 × r(J/  ) Geometrical cross-sections of the J/ ,  ’ and  c states:  geom (J/  ) = 1.96 mb;  geom (  ’) = 7.85 mb;  geom (  c ) = 4.42 mb NA50 data:  abs (  ’) = 7.7 ± 0.9 mb or  abs (  ’/DY) = 8.3 ± 0.9 mb c o i n c i d e n c e ? (no nuclear effects considered here)

15. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 15 Feed down influence on J/  absorption We can redo the Glauber calculations assuming 60% / 30% / 10% as the fractions of direct J/  production and feed downs from  c and  ’ decays... And fixing the  abs of each of the three states to their geometrical values The result is perfectly equivalent to a fit with a free effective  abs (J/  )  2 /ndf = 1.0 c o i n c i d e n c e ? [Figure made by G. Borges] (no nuclear effects considered here)

16. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 16 Let’s express the  abs values of the three charmonium states in terms of the value of the directly produced 1S state, called “J” to distinguish it from the observed “J/  ” (affected by feed down), assuming that they scale with the square of their radii:  with r(  c ) / r(J) = 1.44 ; r(  ’) / r(J) = 1.8 [values from H. Satz] This is a guess… but it is better to assume an answer based on an educated guess than to ignore the existence of the question… The generic survival probability for the state J (or  ’, or  c ) is then given by: And assuming 60% J, 30%  c feed down and 10%  ’ feed down, the survival probability to be compared with the J/  data is: Now we can fit the existing J/  and  ’ data with a single free parameter: From qualitative hints to more detailed calculations

17. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 17 The charmonium production cross section in p-A collisions fraction of charm-anticharm cross section below 2m D K factor to match the magnitude of the LO and NLO cross sections survival probability for nuclear absorption parton density in the proton; j = g, q, qbar parton density in the nucleus nucleon density in the nucleus modification of the parton densities in the nucleus parton momentum fractions F J K th S A abs f j p (x 1,Q 2 ) F j A (x 2,Q 2,b,z’)  A (b,z’) R j (A,x 2,Q 2 ) x 1, x 2 The calculations were done with several PDF sets and nuclear effects models:  GRV LO 94, GRV LO 98, CTEQ6L, MRST2001LO;  non-modified and modified by EKS98, nDS, nDSg, etc

18. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 18 From the measurements of NA3, E866 and HERA-B, respectively at 200, 800 and 920 GeV, we calculated the mid-rapidity ratios between the heavy and light targets of J/  (and  ’) “per nucleon cross sections” 200 GeV : p-Pt / pp = 0.737  0.026 for the J/  800 GeV : p-W / p-Be = 0.8713  0.0263 for the J/  and 0.8032  0.0274 for the  ’ 920 GeV : p-W / p-C = 0.903  0.031 for the J/  Existing J/  and  ’ cross sections in p-A collisions From the NA50 measurements, at 400 and 450 GeV, we calculated the J/  and  ’ cross section ratios, between the heavy targets (Al, Cu, Ag, W, Pb) and Be 400 J/ ’’ Al/Be 0.936  0.0291.015  0.071 Cu/Be 0.907  0.0280.829  0.063 Ag/Be 0.847  0.0260.776  0.060 W/Be 0.804  0.0250.655  0.052 Pb/Be 0.788  0.0240.660  0.049 450 J/ ’’ Al/Be 0.965  0.0410.967  0.051 Cu/Be 0.934  0.0370.893  0.044 Ag/Be 0.866  0.0340.799  0.038 W/Be 0.792  0.0360.706  0.039 2 data points 1 data point 10 data points 8 data points

19. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 19 Extraction of the absorption cross section For each energy and target, the calculations were made with several N-PDFs and for  abs values between 0.0 and 8.0 mb, in steps of 0.5 mb Comparing the calculations with the data we derive the “best”  abs and its error  abs (mb) 4.88  0.29 EKS98 3.75  0.27 nDSg 3.39  0.26 none 200 GeV  abs : insensitive to the PDF set but very dependent on the nuclear effects model EKS98

20. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 20 Using GRV LO 94 and no nuclear effects, we get  abs (J)  3.34  0.25 mb   abs (  ’)  (0.45/0.25) 2   abs (J)  10.8  0.8 mb NA50 obtained  abs (  ’)  10.0  1.5 mb   abs (J/  ) ≈ [ 0.6 + 0.3  (0.36/0.25) 2 + 0.1  (0.45/0.25) 2 ]   abs (J)   5.2  0.4 mb NA50 obtained  abs (J/  )  4.6  0.6 mb Conclusion: if we use the same inputs as NA50, we get the same values (with a smaller error because we make a global fit of the J/  and  ’ data points with one single free parameter, while NA50 made two independent fits) Our calculations vs. the NA50 values

21. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 21 (open circles) (closed circles) Significant drop of  abs with collision energy The J/  and  ’ absorption, at mid-rapidity, weakens with increasing collision energy Exp. E lab (GeV) NA3200 NA50400 NA50450 E866800 E920920 (maximum c.m.s. energy of the J/  – N collision) Assuming a power law function, we can extrapolate  abs to lower and higher energies 158 GeV  abs at 158 GeV is ~50–60% higher than at 400–450 GeV !

22. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 22 R dAu Mid-rapidity data point : R(dAu) = 0.84  0.20 (stat. and syst. errors added in quadrature) Cold nuclear matter effects at RHIC energies PHENIX

23. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 23 Extraction of  abs from the PHENIX mid-rapidity d-Au data The calculations were redone with several N-PDFs and for  abs values between 0.0 and 8.0 mb, in steps of 0.5 mb None / nDSg

24. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 24  abs extrapolated to RHIC energies PHENIX (EKS98) The extrapolation from fixed-target energies matches well the PHENIX d-Au data Much more accurate RHIC data needed to verify the functional form of the energy dependence

25. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 25 The  ’ suppression pattern shows a significant and abrupt drop between the “normal extrapolation” of the 450 GeV p-A data and the S-U / Pb-Pb patterns But this “step” happens between data sets collected at very different energies… and will disappear if the  ’  abs increases significantly from 450 to 158 GeV ! NA60 p-A data at 158 GeV will soon address (and hopefully answer) this question ’’ Preliminary ’’ Effect on the  ’ “suppression”: magnitude and shape  abs = 8.3 ± 0.9 mb

26. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 26 The relative comparison between the In-In and Pb-Pb J/  suppression patterns will not change, because they were both taken at the same energy, 158 GeV, precisely to minimise the number of “free parameters” in their comparison But there will be a common decrease of the magnitude and of the slope’s steepness Effect on the J/  “suppression”: magnitude and shape

27. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 27 In the QGP phase the heavy quarkonium states are “dissolved”, at successive temperature thresholds The feed-down from higher states leads to a “step-wise” J/  suppression pattern Quarkonium melting by QGP : thresholds  steps ’’ cc J/  cocktail (in pp): ~ 60–65% direct J/  ~ 25–30% from  c decays ~ 10% from  ’ decays

28. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 28 Step at N part = 86 ± 8 A1 = 0.98 ± 0.02 A2 = 0.84 ± 0.01  2 / ndf = 0.75 (ndf = 8  3 = 5) Taking into account the E ZDC resolution, the measured pattern is perfectly compatible with a step function in N part N part Measured / Expected 1 Step position A1 A2 Maybe there is even a hint of charm “coalescence” in the most central collisions The In-In J/  suppression pattern versus a step function

29. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 29 N part is convenient to compare In-In and Pb-Pb data: derived from the same E ZDC using the same Glauber formalism (except for the nuclear density functions) If the “real variable” driving charmonium suppression is not N part the measured smearing is the convolution of the detector resolution with the “physics smearing” The detector resolution is 20 (in N part ), while a fit to the measured pattern gives 19:  the “physics smearing” is negligible with respect to the ZDC resolution… Is the step in Npart or in another variable? The In-In data indicates a step in the J/  suppression pattern and suggests that “the physics variable” is N part or a variable very strongly correlated to N part

30. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 30 Steps: N part = 90 ± 5 and 247 ± 19 A1 = 0.96 ± 0.02 A2 = 0.84 ± 0.01 A3 = 0.63 ± 0.03  2 / ndf = 0.72 (ndf = 16  5 = 11) N part Measured / Expected 1 Step positions A1 A2 A3 If we try fitting the In-In and Pb-Pb data with one single step we get  2 /ndf = 5 !  the Pb-Pb points rule out the single-step function and indicate a second step What about the Pb-Pb suppression pattern? -12% :  ’ ? -21% :  c ? We urgently need a much more accurate Pb-Pb pattern

31. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 31 Summary and outlook The J/  and  ’ final state “normal nuclear absorption”, determined by p-A data, is insensitive to the PDF set used but is significantly affected by the nuclear effects model assumed: none, EKS98, nDSg, etc  The latest PDF sets are mature, constrained by a wealth of data (DIS, DY, etc), while the nuclear effects on the gluon densities have not yet been measured…  We must measure the open charm nuclear dependence, versus p T and y All existing J/  and  ’ p-A data can be described by Glauber calculations using one single  abs parameter, with the  c and  ’ values fixed by geometrical scaling and the “observed J/  ” value fixed from the feed down fractions  The fitted  abs values show a significant decrease from 200 to 920 GeV ; the value extrapolated to 158 GeV is 50–60% larger than previously assumed…  We must re-evaluate the SPS J/  and  ’ suppression patterns, with N-PDFs and the increased  abs : the “anomalous suppression” will decrease in magnitude and will become less steep

32. Carlos Lourenço — Hot and Dense Matter — Mumbai Feb. 2008 32 Acknowledgements: This work started after discussions with Helmut Satz and Bob Thews, in May 2006 Previous related work: - Hard Probes 2006 talk, by CL - Quark Matter 2006 talk, by RV References of the data points: NA3, NA50, E866 : published papers HERA-B : values reported at HP06 PHENIX : preprint arXiv: 0711.3917 Work done in collaboration with: Ramona Vogt and Hermine Wöhri