Early times and thermalization in heavy ion collisions: a summary of experimental results for photons, light vector mesons, open and hidden heavy flavors
EEarly times and thermalization in heavy ion collisions: a summaryof experimental results for photons, light vector mesons, open andhidden heavy flavors
Hugo Pereira Da Costa
IRFU / SPhN, CEA Saclay, F-91191, Gif-sur-Yvette, France
Abstract
This contribution summarizes the main experimental results presented at the 2009 Quark Matterconference concerning single and dilepton production in proton and heavy ion collisions at highenergy. The dilepton invariant mass spectrum has been measured over a range that extends fromthe π mass to the Υ mass, and for various collision energies at SPS, Fermilab, Hera and RHIC.This paper focuses on the various contributions (photons, low mass vector mesons, open andhidden heavy flavors) to this spectrum and discuss their implications on our understanding of thematter formed in heavy ion collisions.
1. Introduction
Single and dilepton probes in heavy ion collisions are of particular interest since such probes,once produced, are largely una ff ected by the surrounding QCD medium. They carry valuableinformation on the particle from which they originate and allow one to assess the properties ofthe medium formed in the early instants of the collision. The following contributions to thedilepton invariant mass spectrum are discussed here, together with what one might learn fromtheir measurement about the properties of the medium formed in the collision: • Low mass dileptons originating from vector meson leptonic decay ( ρ , φ and ω ) provideinsight on the properties of these mesons in the high temperature expanding fireball pro-duced immediately after the collision, where chiral symmetry may be (at least partially)restored [1, 2, 3]; • A significant fraction of the virtual and direct photons produced at low p T ( p T < / c)in heavy ion collisions originates from the thermal black-body radiation of the createdfireball [4, 5]. Measuring these photons therefore allows one to quantify the temperatureof the fireball; • Open heavy flavors, because of their high mass, allow one to study in-medium energy lossmechanisms in addition to what can be learned from light quarks [6, 7]; • Heavy quarkonia are of interest because of additional mechanisms that are predicted tooccur in the presence of a QGP and that would a ff ect the production of these boundstates [8, 9, 10]. Preprint submitted to Nuclear Physics A October 29, 2018 a r X i v : . [ nu c l - e x ] S e p . Low mass vector mesons Fig. 1 (left) shows the correlated dimuons invariant mass distribution at the ρ vacuum mass,measured by the NA60 experiment in semi-central In + In collisions [11]. The ρ mass peak di ff erssignificantly from the expected vacuum ρ and can be reasonably well described on the low massside by the model presented in [3, 12]. This model includes a detailed description of the baryonicmatter created in the collision below the formation temperature of a QGP, T c . Interactions withthis baryonic matter are responsible for a broadening of the ρ (but no modification of its mass)when approaching chiral symmetry restoration near T c . Figure 1: Left: dimuon invariant mass distribution in the ρ mass region in In + In semi-central collisions measured byNA60 at SPS. Right: low mass dielectron invariant mass distribution in Au + Au collisions measured by PHENIX atRHIC.
A measurement of dilepton invariant mass distributions in the same mass region has been car-ried out by the PHENIX collaboration at RHIC in Au + Au collisions at √ s NN =
200 GeV [13].An excess over expected background sources is observed between 0 . . / c whichcannot be described by models similar to the one above [12], although such models work reason-ably well for larger masses (Fig. 1, right). This low mass excess is larger for low p T dileptons. Apossible contribution to this excess, which has not been accounted for in the calculations above,might come from quark-gluon scattering into a quark and a virtual photon ( qg → q γ ∗ ). A similarcalculation valid for the direct photon production yields at RHIC has been carried out in [14],which accounts for q + g scattering using a complete leading-order QGP emission rate [15]. Thepredicted integrated magnitude of this contribution is about one third of the hadron gas thermalradiation contribution. Applying this to the virtual photon case might explain part of the excessobserved at RHIC, but a detailed calculation is still to be carried out.
3. Direct photons
Direct photon production yields (as a function of p T ) can be derived from the dilepton in-variant mass spectrum using the following steps [16]: 1) consider the excess of dileptons overexpected hadronic sources in the kinematic range m ∈ [0 . , .
3] GeV / c and p T > / c,2here the contribution of low mass vector mesons should be negligible (Fig. 2, left and centerpanels), 2) interpret this excess as a direct virtual photon signal (with photons decaying into di-electrons) and 3) extrapolate this signal to an invariant mass m = Figure 2: Left and center: dilepton invariant mass distribution as a function of mass for di ff erent p T bins, in p + p collisions (left) and Au + Au minimum bias collisions (center). Data are compared to expected background sources toderive a possible virtual photon excess. Right: calculated direct photon yield as a function of p T in di ff erent centralitybins, compared to binary scaled p + p yields. The resulting yields (as a function or p T ) are compared to yields obtained in p + p collisionsscaled by N col , the number of nucleon-nucleon collisions equivalent to one A + A collision in agiven centrality bin, and the di ff erence is fitted to extract a time averaged (over the medium ex-pansion history) black body radiation temperature (Fig. 2, right). For central Au + Au collisions atRHIC energy, a temperature of 221 ±
23 MeV is obtained [17]. These thermal photon yields canalso be compared to various theoretical models in order to derive a medium initial temperature,by making assumptions on how this medium expands and cools down over time [18]. Depend-ing on how long it takes for the system to thermalize, an initial temperature between 300 and600 MeV is obtained. As one might expect, later thermalization times led to smaller initial tem-peratures. Similar fits applied to the Pb + Pb WA98 direct photon measurements [19] give aninitial temperature of about 200 MeV [20].
4. Open heavy flavor
There is still a disagreement of about a factor two between the STAR and PHENIX heavyflavor (charm and beauty) total cross-section measurements in p + p , d + A and A + A collisionsat √ s =
200 GeV [21, 22], as well as between the open charm di ff erential cross-section asa function of p T [21] (Fig. 3). The main di ff erences between the two experiments are 1) theamount of material in the detector acceptance 2) the rapidity and p T range of the measuredelectrons used for heavy flavor identification. E ff orts are underway in both collaborations tobetter understand existing measurements and provide new independent measurements in order toaddress this discrepancy: • The PHENIX collaboration is working on refining its understanding of the electron cock-tail which is subtracted from the raw single electron spectrum to derive the heavy-flavor3 igure 3: Left: total heavy-flavor production cross-section as a function of N col measured by PHENIX and STAR atRHIC in p + p , d + A and A + A collisions. Right: ratio between the heavy-flavor di ff erential production cross-sectionas a function of decay electron p T measured by PHENIX and STAR in p + p collisions and a FONLL calculation. signal, and now accounts for the contribution of electrons coming from J /ψ , Υ and Drell-Yan [23]. PHENIX also measured the total D + B production cross-section in a largely inde-pendent way by estimating all the contributions to the dielectron invariant mass spectrum(as opposed to the single electron spectrum) using data-driven simulations [24]. FinallyPHENIX reported on a first study of electron-muon correlations to measure DD productionin a way that is largely free of background [25]; • The STAR collaboration has removed its central silicon detector in order to reduce theamount of material in the spectrometer and the corresponding photo-conversion back-ground contribution to the raw single electron spectrum. It also measured the productionof low p T D mesons using their decay into a K , π pair, and using single muons [26, 27];In A + A collisions, the measurement of the heavy flavor nuclear modification factor R AA agrees between the two collaborations [21, 22]. The heavy flavor production at high p T ( p T > / c) exhibits a large suppression with respect to binary scaled cross-sections in p + p (Fig. 4,top-left). This indicates that high p T heavy quarks lose a significant fraction of their energy whentraversing the medium created during the collision, and poses a challenge to theoretical models,since heavy quarks, due to their high mass, are expected to loose less energy (via gluon radiation)than light quarks [28]. Additionally, a large elliptic flow v is observed for intermediate p T heavyquarks (1 < p T < / c) in Au + Au minimum bias collisions (Fig. 4, bottom-left), indicatingthat intermediate p T heavy quarks are rapidly thermalized. These two observations are inter-preted as an evidence for a strong coupling between the heavy quarks and the medium producedduring the collision. No consensus amongst theorists has been achieved to date concerning theunderlying mechanism responsible for this strong coupling (see e.g. [29, 30, 31].Current single lepton measurements do not allow for a separation of charm and beauty in amodel independent way. However, separate measurements have been performed to determinethe relative contribution of charm and beauty to total heavy flavor yields. These are either di-rect measurements (using the hadronic decay of D mesons), or indirect measurements (e.g. bystudying the correlation of opposite sign electron-hadron pairs in the final state to separate thecontributions of D and B semi-leptonic decays). In p + p collisions, the resulting B / (B + D) ratiosagree well between STAR and PHENIX [32]. They are consistent with a Fixed Order Next toLeading Log (FONLL) calculation [33] (Fig. 4, right).Measuring the total heavy flavor R AA and the B / (B + D) ratio in p + p collisions allows oneto uniquely relate the R AA of B and D mesons: smaller values of R DAA bring R BAA closer to unity.The (negative) slope of the relation between the two is driven by the D / B ratio measured in p + p igure 4: Left: Heavy-flavor electron R AA and elliptic flow measured by PHENIX in Au + Au collisions at RHIC; right:B / D + B production ratio as a function of p T in p + p collisions measured by STAR, compared to FONLL calculations. collisions whereas its magnitude is controlled by the total heavy flavor R AA . The main conclusionof such an analysis [32] is that even in the unlikely case where high p T charm quarks are entirelysuppressed in A + A collisions, a significant suppression of high p T b quarks is still needed toexplain the total heavy flavor R AA measured at RHIC. This poses an even greater challenge totheoretical models than the charm R AA , since b quarks are significantly heavier than c quarks.More information will be gained on this matter by measuring charm and beauty separately inA + A collisions. Both STAR and PHENIX are undergoing silicon vertex detector upgrades forthe central tracking that should allow direct measurement of D and B mesons.
5. Heavy Quarkonia
Heavy quarkonia have been studied extensively at the SPS and the RHIC since they arepredicted to melt, via QCD Debye screening, in the presence of a Quark-Gluon Plasma [8].Recently, focus has been given to understanding both the heavy quarkonia production mecha-nism in p + p collisions and the cold nuclear matter e ff ects which a ff ect the production of heavyquarkonia when colliding two nuclei without the formation of a QGP.Heavy quarkonia production yields in p + p collisions serve as a reference to study mediume ff ects in p + A, d + A and A + A collisions and help in understanding how these bound statesare produced. Fig. 5 (left) shows the J /ψ production invariant yields as a function of rapiditymeasured in p + p collisions at RHIC by PHENIX using the 2006 high statistics p + p datasample [34]. These yields can be compared to calculations that assume di ff erent underlyingproduction mechanisms, however both statistical and systematic uncertainties are still too largeto uniquely identify the correct mechanism at play. Another way to address the productionmechanism is to measure the J /ψ polarization since models have very di ff erent predictions forthis observable. Fig. 5 (right) shows the J /ψ polarization measured in the helicity frame byPHENIX in p + p collisions at mid and forward rapidity [34]. The model shown on the figure(a refined version of the Color Singlet Model [35]) reproduces reasonably well the data at mid-rapidity but misses the measurement at forward rapidity. Similarly, all available measurements onJ /ψ polarization have been collected, rotated so that they are all evaluated in the same referenceframe (here the Collin-Sopper frame [36]) and represented as a function of the J /ψ momentum. Aglobal trend is observed that is largely independent of the collision energy but lacks a theoreticalexplanation [37]. 5 igure 5: Left: J /ψ production yield as a function of J /ψ rapidity measured in p + p collisions at RHIC. Right: J /ψ polarization measured in the helicity frame by PHENIX in p + p collisions. Cold nuclear matter e ff ects must be carefully evaluated and properly accounted for whenconsidering yield modifications observed in A + A collisions before quantifying the e ff ects of aQGP. They include: modification of the parton distribution functions (pdf) in the nucleus (no-tably shadowing or gluon saturation at low x Bj , anti-shadowing at large x Bj ); nuclear absorp-tion / dissociation; initial state energy loss and the Cronin e ff ect. The general approach used upto now to quantify the cold nuclear matter e ff ects [38] is to choose a set of modified pdfs, addsome e ff ective absorption (or break-up) cross-section to account for the other possible e ff ects,derive the resulting expected heavy quarkonia production yield, and fit this expected yield to the p + A or d + A available measurements, leaving the e ff ective break-up cross-section as a freeparameter. These e ff ects are then extrapolated to A + A collisions and compared to the data.At the SPS, an updated break-up cross-section has been estimated that properly accounts forthe fact that the gluon x domain covered by the experiments corresponds to the anti-shadowingregion of modified pdfs, for which the gluon content is enhanced with respect to the bare nucleoncase (see e.g. [39]). Consequently, the new cross-section derived from p + A data is significantlylarger than the previously published value. When extrapolated to In + In, the J /ψ suppressionfactor estimated from cold nuclear matter e ff ects matches the data rather well and leaves littleroom for any additional anomalous suppression [40] (Fig. 6, left).At RHIC, updated break-up cross-sections have been derived from the new 2009 d + Au datasample which is about 30 times as large as the one used for previous published results [34]. Thesecross-sections must still be extrapolated to Au + Au collisions in order to quantify any additionalanomalous suppression due to the possible formation of a QGP.A systematic survey of the e ff ective charmonia break-up cross-section has been performedthat collects results from SPS, HERA, Fermilab and RHIC [41]. When plotted as a function of thecollision energy a common (exponentially decreasing) trend is observed although this trend hasno theoretical interpretation yet (Fig. 6, right). When represented as a function of rapidity, anddisregarding the collision energy, the e ff ective break-up cross-section also exhibits a somewhatuniversal trend, that cannot be easily explained in terms of the e ff ects listed above. Note thatsimilar surveys have been performed in the past that led to di ff erent conclusions, namely that thecurrent data are consistent with no energy dependency [42].The first Υ measurements have become available at RHIC (with limited statistics) in p + p , d + Au and Au + Au collisions (Fig. 7). Due to limited statistics, it is di ffi cult to disentangle the Υ signal and the underlying correlated background sources (from Drell-Yan and open beauty). One6 igure 6: Left: J /ψ R AA at SPS after removal of CNM e ff ects measured by NA60. Right: J /ψ e ff ective break-upcross-section as a function of collision energy in d + A or p + A collisions. can either ignore these contributions and derive e.g. nuclear modification factors for inclusivehigh-mass dileptons, or estimate them from simulations and use the corresponding uncertainty asa systematic error. In p + p collisions a total Υ production cross-section BRd σ/ dy ( | y | < . = + − pb is measured [34]; in d + Au collisions a nuclear modification factor consistent withunity is observed [43] while in Au + Au collisions this nuclear modification factor is smaller that0.64 at 90 % confidence level [44], meaning that inclusive high mass dileptons are significantlysuppressed by the medium formed in Au + Au collisions at √ s NN =
200 GeV.
Figure 7: dielectron invariant mass distributions at high mass in p + p (left), d + Au (center) and Au + Au collisions,measured by PHENIX and STAR at RHIC.
6. Conclusion
In short: • Low mass vector mesons exhibit strong shape modifications with respect to their vacuumproperties, that can be well described at SPS but not at RHIC possibly because somecontributions to the dilepton spectrum have not been properly accounted for; • Virtual photons can be used in addition to direct photon measurements to assess the mediumtemperature averaged over its expansion time and derive its initial temperature; • A significant suppression of b quarks is necessary to describe the observed heavy flavor R AA in a way that is consistent with the B / B + D ratio measured in p + p collisions;7 J /ψ production in heavy-ion collisions is a puzzle. The situation is more complex than theoriginal picture, due to our poor knowledge of its production mechanism in p + p collisionsand to the existence of many cold nuclear matter e ff ects which significantly modify thisproduction even in the absence of a QGP. E ff orts are being made to better understand theabove so that one can quantify the hot , abnormal e ff ects at both SPS and RHIC. Notably, itappears that the suppression measured at SPS in In-In collisions can be entirely describedin terms of such cold nuclear matter e ff ects. References [1] Pisarski R D 1982,
Phys. Lett.
B 155[2] Brown G E and Rho M 2002,
Phys. Rep.
Adv. Nucl. Phys. Ann. Rev. Nucl. Part. Sci. et al. Phys. Rev. C ff D and Zakharov B G 2000,
Annu. Rev. Nucl. Part. Sci. et al. nucl-th / Phys. Lett. B et al. Phys. Lett. B Nucl. Phys. A et al. (NA60 collaboration) 2006, Phys. Rev. Lett. / et al. (PHENIX collaboration) 2007, arXiv: 0706.3034v1 [nucl-ex][14] Turbide S, Rapp R and Gale 2004, Phys. Rev. C ff e L G 2001, JHEP et al. (PHENIX collaboration), arXiv: 0804.4168v1 [nucl-ex][18] d’Enterria D and Peressounko D 2006,
Eur. Phys. J. C et al. (WA98 collaboration) 2000, Phys. Rev. Lett. phys. Rev. C et al. (STAR collaboration) 2007 Phys. Rev. Lett. et al. (PHENIX collaboration) 2007, Phys. Rev. Lett et al. (PHENIX collaboration) 2009, Phys Lett. B et al. (STAR collaboration) 2005, Phys. Rev. Lett. et al. (STAR collaboration), arXiv:0805.0364[28] Dokshitzer Y L and Kharzeev D E 2001, Phys. Lett. B et al. Phys. Lett. B Phys. Rev. C Phys. Rev. C et al. Phys. Rev. Lett phys. Rev. Lett Phys. Rev. D et al. (PHENIX collaboration) 2008, Phys. Rev. C Eur. Phys. J C449[43] Liu H (STAR collaboration), this proceedings[44] Linden Levy L A, this proceedings