Ceres measures lepton pairs

Dilepton-enhancement in relativistic heavy-ion collisions

The CERES- experiment at the Cern-SPS investigates dielectrons in relativistic heavy-ion collisions in the region of low invariant masses. In and below the mass region of the rho-meson the dilepton rates in medium- and very heavy systems are enhanced as compared to the expectation from hadronic decays, whereas no deviations are found for proton-induced reactions. A possible origin is the change of the vector-meson properties in hot and dense matter. Mass resolution and statistics of the experiment are further improved.

Leptonic probes play a special role in the search for possible signatures of quark-gluon plasma formation in relativistic heavy-ion collisions. Due to their electroweak - not strong - interaction, they transport information about the state of the strongly interacting system at the time of their creation directly to the detector.

In a thermalized plasma phase, for example, an interaction of quark and antiquark would produce a virtual photon, which then decays in a lepton and an antilepton - a dilepton - with a certain invariant mass. However, one has not yet detected such signals in the region of medium and large invariant masses (Fig.2), and in the low-mass region up to 1.5 GeV/c^2 that is relevant for the CERES experiment such processes are completely covered by others - which are, however, interesting enough to be investigated in large-scale experiments. Here, dileptons are mostly created from the interactions of charged hadrons - mostly pions - and their antiparticles, and from the decay of hadron resonances such as rho, omega and phi.

In the Cern experiments HELIOS-1 (p-Be), HELIOS-3 (p-W, S-W) and CERES (p-Be, p-Au, S-Au and Pb-Au) the decay of the vector mesons in dileptons at relatively low invariant masses has proven to be especially important and interesting because it allows to investigate the properties of these short-lived particles in hot and dense matter. As an example, a free rho-meson has a mass of 770 MeV and a mean lifetime of 4.4*10^-24 seconds. Both values can be changed due to strong interaction in the hot and dense medium. Any change of mass or width of the rho-meson (any modification of the real or the imaginary part of the in-medium spectral function) modifies the invariant mass spectrum of dileptons that are produced in the decay of the meson. The analyses of Ceres-results give clear indications that this is indeed the case. It should be noted, however, that only a tiny fraction of about 1 in 10^4 rhos decay electroweakly in lepton pairs via a virtual photon; most of the decays are in pions and are not registered in Ceres.

In contrast, the much heavier (3.1 GeV) J/Psi- meson (Fig.2) usually decays in vacuum due to its long lifetime of about 10^-20 seconds, such that dileptons from its decay do not carry information about in-medium properties. Instead one can learn something about J/Psi production rates, and possible destructions due to final-state interactions, through dilepton measurements. This is being investigated in another fixed-target experiment at Cern (NA50) that complements Ceres in an ideal way at higher invariant masses. NA50 has indeed found some indications for a possible quark-gluon plasma formation through suppressed J/Psi-rates in very heavy systems (see page ... of this issue).

In this region the Drell-Yan process is the dominat source of the dilepton-continuum: a valence quark in the nucleon of one nucleus interacts with a (sea-) antiquark in a nucleon of the other nucleus; annihilation generates a virtual photon, which decays in a lepton pair (Drell-Yan in fig.2). At low masses, however, this process and also dileptons from created charm (DDbar) are not important, such that in the Ceres-experiment the dileptons are essentially due to the decay of neutral mesons. The peak structure results from the direct rho- and omega-decays which are usually not resolved, and from the phi-meson at 1.02 GeV, whereas the Dalitz-decays of the neutral mesons pi, eta, eta' and omega generate dilepton-continua. The cross section increases strongly at low masses and small transverse momenta.

The spectrometer

The crucial experimental problem in relativistic heavy-ion dilepton-experiments at SPS-energies stems from the fact that the number of charged hadrons exceeds the one of dileptons by a factor of 10^5. Hence, one has to choose a detector system which is essentially insensitive to hadrons.

The Ceres-collaboration - scientists from GSI, Weizmann institute, Heidelberg University and seven other institutes - has developed a CERES (ChErenkov Ring Electron Spectrometer) for low invariant masses [1]. It is based on two RICH (Ring-Imaging Cerenkov) detectors and a radial drift chamber (Fig.1) which cover a rapidity region of 2.1 < y < 2.7. In the medium leptons travel faster than light, generating a Cerenkov-lightcone that resembles a supersonic shock front in case of sound waves. The ring-images of the cones are then used to detect the leptons.

The detectors operate at a threshold of gamma = 32 and hence, they are not sensitive for about 95 % of all charged hadrons, whereas dielectrons can be detected with good resolution and statistics. Pions with high momentum can be detected, but it is possible to reject them due to the significantly smaller radii of their Cerenkov-lightcones. This offers also a possibility to measure charged high-momentum pions. Momenta are determined through the azimuthal deflection of the charged particles in a magnetic field between the two RICH detectors.

The RICH-detectors are complemented by two Silicon-driftdetectors (SDD1,2 in fig.1) 10 centimeters behind the target, and by a multiwire-proportionalchamber (MWPC) at the end of the spectrometer. They were installed as an upgrade for the Pb-runs in order to provide a more accurate reconstruction of the physical sources of registered signals: in a Pb-Au experiment, up to 300 charged particles are produced in the CERES acceptance region.

A significant problem is the very large number of photons per electron-positron pair of about 10^5. Hence, the radiation length within the detector acceptance region has been reduced such that only 1 % of all real photons convert to an electron-positron pair in this region. Still, in the mass region below 200 MeV/c^2 this remains the dominant dielectron source, together with the decay pi0 in e+e-gamma. As a consequence, Ceres can successfully identify other dilepton sources only above 200 MeV/c^2. On the other hand, the conversion of real photons in a 100 micrometer steel foil as an additional photon-converter can be used to measure photon spectra.

An important problem in the data analysis arises from the limited acceptance and signal detection efficiency. Two only partially reconstructed dilepton-events at low mass can generate a spurious signal at higher invariant mass. In oder to reduce this combinatorial background, especially the detection of dileptons at low mass has to be improved. Therefore the magnetic field is designed in such a way that the path leading to the first Cerenkov-detector RICH1 is field-free, such that the small opening angle of the pairs can be measured here with RICH1 and the two drift detectors, in order to be used later to reduce the combinatorial background. Pairs with small opening angles can thus also be identified in case the trajectory of one of the particles is measured incompletely. As an additional check, a monte-carlo generated combinatorial background agrees with the background of the data within 3 %.

The experiment was proposed in the middle of 1988 and approved in early 1989. First data were taken in the beginning of 1992. In order to investigate dielectrons from proton-induced collisions, the electromagnetic calorimeter TAPS was added in 1993. It consists of BaF2 crystals and covers a rapidity region of 3.1 < y < 4.0. CERES/ NA45 is the only dielectron-spectrometer among the SPS-experiments. The NA50 setup - which is designed to cover the higher mass region of 1.5 to 7 GeV/c^2 - rejects charged hadrons using absorbers, and measures myonpairs in large-volume chambers.

Enhanced dilepton-rates

In the 1992 beamtime Ceres detected significant deviations of the dielectron-data from the expectation that was based on the decay of neutral mesons only, fig.3 [1]. With the exception of the first three data points which were well represented through Dalitz-decays of neutral pions, the data are consistently above the expectation (solid curve). Moreover, the shape above twice the pion mass deviates significantly: the data have a local maximum around 0.4 GeV/c^2, where a minimum is expected. The minimum had been "filled up", and the rho/omega-resonance is covered by this plateau. Integrating the data in the mass region from 0.2 to 1.5 GeV/c^2, the result from neutral-meson decays is exceeded by a factor of 5, with statistical uncertainties of 0.7 and systematic uncertainties of 2. The SPS-experiments HELIOS-3 and NA38 also found significantly enhanced dilepton rates (dimyons) in collisions of sulphur with heavy targets (bismuth and uranium). In all of these cases, the enhancement is most pronounced in the mass region of 300 to 700 MeV/c^2.

In these comparisons between data and a calculated dilepton-"cocktail" from neutral-meson decays, the latter is based on data from proton-induced collisions which were scaled with the multiplicity of charged particles. The transverse momentum and rapidity distributions were adapted to the conditions of the nucleus-nucleus collision, the acceptance of the experiment, and the conditions of the off-line data analysis were considered.

These observations started an intense discussion about the processes leading to dilepton-emission from hot and dense hadron- or quark-matter. It soon turned out that dileptons from pion-annihilation and scattering processes are not sufficient to explain the enhancement. Many groups performed a multitude of model calculations based on microscopic transport or hydrodynamic codes, and their results agreed within a certain band width: There is apparently no strong dependence on the details of the dynamical descriptions. Although the results of the calculations were much closer to the data in the region of and beyond the rho-mass than the cocktail of dileptons from freely decaying hadrons, none of the model results could reproduce the relative maximum near 400 MeV/c^2. Hence, it seemed possible that the data offer an indication for "new physics", namely, a modification of the properties of vector mesons in hot and dense matter.

Proton-induced reactions

In view of the great importance of this result, it turned out to be necessary to reinvestigate proton-induced collisions, and to compare the results with the model calculations. The Ceres- and TAPS-groups have studied dileptons from p-Be and p-Au collisions @ 45o GeV/c in the region of low invariant masses up to 1.5 GeV/c^2 (Fig.4). Dielectrons were detected in coincidence with photons in order to obtain precise data about the production of neutral mesons at midrapidity, and to reduce the uncertainties in the determination of hadronic sources in the dilepton-spectra [2].

The Dalitz-decays of neutral pi- and eta-mesons were completely reconstructed through the detection of the dielectrons in the Cerenkov-spectrometer, and of the photon in the electromagnetic calorimeter of TAPS. Precise data for pi-, eta- and omega-mesons were obtained from their gamma-gamma and pi-gamma decay modes.

For p-Be 5068 dileptons with invariant masses larger than 200 MeV/c^2 were analyzed, for p-Au, 1068. As expected from the Dalitz-decays of pi- and eta-mesons, the spectra decay rapidly with rising mass (Fig.4). The structure from the decay of the rho/eta-mesons at about 780 MeV/c^2 is clearly visible. From the measured width of this maximum the mass resolution is about 11 % in agreement with the expectation from the momentum resolution of CERES.

The inclusive dilepton-spectra are shown in fig.4 together with the contributions from individual hadron decays, and their sums (solid curves). With rising mass, the dilepton rates drop by five orders of magnitude. For 100% acceptance of the spectrometer for lepton pairs they would be less steep, but in view of the desired direct comparison with heavy-ion data a correction for pair acceptance was not made.

The comparison between data and hadrondecay-cocktail shows very good agreement within the systematical uncertainties. For the beryllium-target, earlier results of the Helios-1-collaboration were confirmed, with improved systematical uncertainties 23% in the mass region below 450 MeV/c^2). Any non-conventional source for dileptons is therefore limited to several percent of the hadron-decay contributions. In the search for new phenomena in heavy-ion induced collisions (Fig.3) these data were an important point of reference; in particular, the analysis of the beryllium-data was used in the scaling of conventional dilepton-sources in the S-Au system, which resulted in the recognition of the dilepton-enhancement in heavy systems.

Heavy systems

When the 158 A GeV/c lead-beam became available at the SPS in 1995, the search for an enhancement of dilepton production at low invariant masses in very heavy systems was one of the main research goals. CERES had been upgraded by two silicon-driftdetectors and a multiwire-proportional chamber in order to cope with the high multiplicities in Pb-Au collisions. Together with the RICH-data, the required efficiency of the signal reconstruction procedure could be attained.

As in case of the S-Au system, the invariant mass spectrum from meson decays agrees with the Pb-Au data below 200 MeV/c^2 (Fig.6); here, the dilepton rates are dominated by Dalitz-decays of neutral pions. Again, the dilepton-rates are significantly enhanced at higher invariant masses - most pronounced, in the region between 250 and 700 MeV/c^2. Integration yields an enhancement factor of 3.9 above the result of the hadronic cocktail in this region, with statistical uncertainties of 0.9 and systematic uncertainties of 0.9. Here, the relative particle abundancies in the generated spectrum are determined in a thermal model, with parameters (temperature, chemical potential) adjusted to the particle abundancies measured by the NA49-collaboration in the Pb-Pb system.

The enhancement in the data prevails up to the highest masses measured by Ceres, with the exception of the phi-region. The smooth shape of the experimental result has no similarity at all with the spectrum that is generated from hadronic decays; in particular, no rho/omega and phi- resonance structures can be seen in the data. This is qualitatively similar to the former S-Au result.

In an additional beamtime in 1996, these results could be confirmed with much better statistics (black dots in fig.6). However, the data are systematically lower, such that the integral enhancement factor is only 2.6, with statistical uncertainties of 0.5 and systematical uncertainties of 0.6. The similar shape of the dilepton spectrum in sulphur- and lead- induced collisions, and the enhancement as compared to the results from hadronic decays, is clearly visible in fig.7.

In the critical mass region between 250 MeV/c^2 and the rho-mass, the enhancement depends on the centrality of the reaction, fig. 8. Here, the number of detected charged particles per rapidity interval is taken as a measure for the centrality of the collision. Since the expectation from hadron decays rises linearly with multiplicity, a constant enhancement corresponds to a linear dN/dy- dependence. Hence, the enhancement in the region below the rho-mass depends stronger than linearly on the particle multiplicity and thus, on the impact parameter: central collisions lead to a strong enhancement in the dilepton spectra, whereas the enhancement is low in peripheral collisions. Evidently, the effect depends strongly on both baryon and energy density.

Moreover, there is a dependence of the enhancement from the transverse momenta of the lepton pairs: the effect is more pronounced at low transverse momenta (< 500 GeV/c). However, due to systematic discrepancies of the 1995 and 1996 results a detailed analysis is not yet available.

In the 1996 beamtime, CERES investigated also 34000 Photons which converted to dileptons in a converter foil between the two silicon-driftdetectors. Due to the clear signature of such events, the background is negligible. In fig. 10 the inclusive photon spectrum as function of transverse momentum is compared with the expectation from hadronic decays. Both data and generated events are normalized to the number of photons per neutral pion. Apart from small deviations, the data follow the expectation from hadronic decays. In order to further reduce the systematic uncertainties, new measurements in this years beam time are prepared.

Ceres also measured charged hadrons. The events were reconstructed with two independent methods. The result of an indirect method is in good agreement with the measurement of charged pions with momentum > 4.5 GeV/c from their emission of Cerenkov light. The agreement is improved when the contributions of kaons and antiprotons are deducted.

Theoretical interpretations

The enhancement of the dilepton rates in heavy system at low invariant masses is the outstanding result of Ceres that needs theoretical interpretation. As for the S-Au system, the theoretical efforts had rapidly clarified that the enhancement effect cannot be understood solely on the basis of dilepton contributions from pion annihilation, and this was confirmed for the Pb-Au system. Although the annihilation contributions lift the theoretical results significantly above the outcome from hadronic decays and improve the agreement with the data in the region of the rho-mass, they fail to describe the measured plateau in the region around 400 MeV/c^2. As a consequence, modifications of both properties and propagation of the vector mesons in the hot and dense medium turned out to be necessary conditions for a theoretical interpretation of the Ceres data. A quantitative reproduction of the data turns out to be possible with the assumption that the rho- and omega-masses are reduced in dense matter. Such an assumption is supported by scaling properties of hadron masses relying on invariance properties of the QCD-Lagrangian that were proposed by theorists already in 1991, and also by earlier calculations on the basis of QCD- sumrules. However, the sumrules can also be fulfilled without changing the meson masses if one considers instead a broadening of the resonances as a consequence of multiparticle effects. Based on such broadening of the resonances, and momentum dependencies of the meson self energies in nuclear matter, the Ceres data can also be understood. Possible connections between the two lines of thought are a topic of great theoretical interest. In typical models for the broadening of the spectral function of the vector mesons one obtains a width of the rho-meson in normal nuclear matter density (0.16fm^3) of 200-300 MeV, about twice the value in a pure meson gas. In the dense medium created in relativistic heavy-ion collisions, the width is further enhanced and hence, the maxima in the dilepton spectra that correspond to the resonances are smeared out, and satisfactory reproductions of the plateaus in the data are possible. In some cases, transport calculations with very moderat in-medium corrections yield good agreement with the data, fig.9. In this particular example, the price to be paid for a good reproduction of the measured plateau is a high rate of Dalitz-decays of the omega-meson. As a consequence, the data near 0.8 GeV/c^2 are overestimated due to the direct omega-decays. Here and in case of the numerous other calculations one has to wait for future Ceres-results with much better statistics and mass resolution. A topic of particular interest is the possible connection between modifications of the vector-meson properties as function of baryon density and local energy density on the one hand, and the restoration of chiral symmetry in a chiral phase transition on the other hand. Below such a phase boundary, right-handed quarks are coupled to left-handed antiquarks and vice versa, whereas above the phase boundary right-handed quarlks and antiquarks are independent from left-handed ones such that chiral symmetry (symmetry in the handedness) is restored. Lattice gauge calculations - with the assumtion of thermal equilibrium that is of limited validity in collisions, and with the unrealistic assumption of equal quark- and antiquark-abundancies - show that a deconfinement of quarks and gluons in a quark-gluon plasma occurs at nearly the same energy density as a chiral phase transition. However, as in the search for signatures of deconfinement, it is not completely clear how the restoration of chiral symmetry translates into observables. A measure for the breaking of chiral symmetry is the quark condensate. In lattice gauge calculations, its expectation value drops to very small values when the critical temperature is approached. Via QCD-sumrules, the quark condensate is connected to the masses of vector mesons such as the rho. This motivates a linear dropping of vector meson masses with increasing density, and allows for a reproduction of the enhanced dilepton rates which would then be related to the restoration of chiral symmetry, see above. In view of the alternative interpretation through broadening as a consequence of many-particle effects - which yields somewhat different results especially in the region between omega- and phi-mass - it would be to early to draw definite conclusions from the Ceres-data with respect to a possible restoration of chiral symmetry.

The future

In view of the great significance of the Ceres-results, and of ever more detailed theoretical interpretations, it is necessary to improve the precision in the fixed-target experiments at SPS that was attained in the beam times 1995/96 for heavy systems regarding mass resolution, statistics and signal/background relation. Then, precision spectroscopy of the omega- and phi-resonances should be feasible.

Ceres will continue the measurements in the years 1999/2000 with a 1997/98 newly constructed time projection chamber (TPC, fig.5) inside the field of a new magnet at the downstream end of the spectrometer (Fig.1). In addition, the electronic data analysis was improved such that higher rates can be dealt with.

The radial drift TPC has 16 readout-chambers (ROC's) of two meter length and 0.5 meter diameter. The detector has two independent gas volumes: an outer isolating volume filled with CO_2, and an inner counting volume with a 4:1 mixture of Ne and CO_2. In this years' SPS-run, calibration of the TPC - which has proven to agree with its specifications already in 1998 - will be completed, and then the physics runs will start again.

This year, dileptons from the heavy Pb-Au system will be measured at the much lower beam momentum of 40 A GeV/c. At this incident energy,one expects a higher baryon density as a consequence of longer interaction times, whereas the energy density will be lower as in case of the 158 A GeV/c data. This could be of central importance when trying to distinguish the various theoretical interpretations of the dilepton-anomaly. In the year 2000, the Pb-Au system will then again be investigated at 158 GeV/c with the upgraded CERES and it will be interesting to see how the interpretations evolve - also in view of the first results from RHIC, which will then be available.



[1] G. Baur et al., Nucl. Inst. Meth. A343 (1994) 87; G. Agakichiev et al., ((insert titles please)) Phys. Rev. Lett. 75 (1995) 1272

[2] G. Agakichiev et al., Systematic study of low-mass electron pair production in p-Be and p-Au collisions at 450 GeV/c, Eur. Phys. J. C4 (1998)231

[3] G. Agakichiev et al., Low-Mass e+e- pair production in 158 A GeV Pb-Au collisions at the CERN SPS, its dependence on multiplicity and transverse momentum, Phys. Lett. B422 (1998) 405 ((check this reference please))

[4] B. Lenkeit et al., Recent results from Pb-Au collisions at 158 GeV/c per nucleon obtained with the CERES spectrometer, Proc. Quark Matter 1999, Torino, in press

[5] R. Rapp, Duality and chiral restoration from low-mass dileptons at the CERN-SPS, hep-ph/9907342


Fig 1: The CERES (ChErenkov Ring Electron Spectrometer) / NA45- experiment at the Cern-SPS investigates electron-positron pairs at low invariant mass which are produced in relativistic collisions between heavy nuclei. The spectrometer consists of two Ring Imaging Cerenkov (RICH) detectors which are separated by a superconducting magnet for momentum- and charge determination. Two silicon-driftdetectors (SDD1,2) and a multiwire-proportional chamber (MWPC) further support the reconstruction of the physical dilepton sources. In 1997/97 a time projection chamber (TPC) was added, cf. fig.5. Bildquelle:CERES

((wähle "upgraded with TPC")) Fig. 2: Invariant dilepton mass spectrum (schematic) from relativistic heavy-ion collisions with regions of low, medium and high mass. In the low-mass region that is covered by CERES, the direct decays of light vector mesons rho, omega and phi in lepton pairs are of prime interest. One expects information about changes of the vector-meson properties in the hot and dense medium, and about a possible connection to the restoration of chiral symmetry. Bildquelle:spectrum

Fig. 3: Result of the 1992er S+Au run at 200 A GeV/c,where Ceres found significant deviations of the dilepton-rates from the expectation based on the decay of neutral mesons in p-n and p-A reactions: The dilepton-production in and below the mass region of the rho-meson is enhanced as compared to the results based on meson decays (solid curve) Bildquelle:http://ceres6.physi.uni-heidelberg.de/ceres/referenceData/Index.h tml#S+Au


Fig. 4: The 1993 control run with the combined CERES/TAPS- spectrometers yielded a dilepton spectrum for the recation 450 GeV/c p + Au which agreed with the expectation from lepton pairs of decaying neutral hadrons (lower part). A proton-induced reaction with the lighter beryllium target (upper frame) gives similar results, and confirms earlier Helios-1 measurements. Bildquellen:p-AU


Fig. 5: The Ceres/NA45 was upgraded in 1997/98 with an additional magnet and a new detector. The detector shown here is a radial drift timeprojection-chamber (TPC) with 16 readout-chambers. The chamber is 2.5 meters long and has a width of 3 meters.


Fig. 6: Ceres/NA45- dilepton spectra in 158 A GeV/c Pb + Au at the Cern-SPS. The 1996 data have smaller statistical and systematical uncertainties than the 1995 data, and they are systematically lower. As is case of S+Au (Fig.3), the dilepton production in the region of and below the rho-mass is significantly enhanced as compared to the expectation from the decay of neutral mesons. Bildquelle:http://www.qm99.to.infn.it/lenkeit/le9.gif Enhancement in Pb-Au

Fig. 7: Ceres dilepton spectra for S+Au und Pb+Au- collisions at SPS energies consistently show the enhancement as compared to the result expected from hadronic decays, solid curve. Bildquelle:http://www.qm99.to.infn.it/kluberg/kl6.html Enhancement

Fig. 8: The Ceres data depend on the centrality of the reaction: The enhancement factor of the dilepton rates (as compared with the expectation from the decay of neutral mesons) rises in the mass region between twice the pion mass, and the rho/omega mass stronger than linearly with the multiplicity of charged particles. This supports theoretical interpretations of the enhancement as a consequence of dropping effective rho- and omega-masses in the hot and dense medium, or of an increase in their widths due to many-body effects. Bildquelle:http://www.qm99.to.infn.it/lenkeit/le11.gif Centrality dependence

Fig. 9: Ceres-dilepton data in comparison with a typical relativistic transport calculation (RBUU). The enhancement of the data in the mass region around 0.4 GeV is almost reproduced, with only minor modifications of the in-medium properties of the vector mesons. For more precise comparisons with model calculations, mass resolution and statistics of the Ceres data have to be improved.

Bildquelle: R. Rapp, Duality and Chiral Restoration from low-mass dileptons at the Cern-SPS, hep-ph/9907342 (Proc. QM99 Torino), Fig. 8.

Farbbild: ps-file via Johanna Stachel folgt.

Fig.10: Inclusive photon spectrum as function of transverse momentum in the reaction Pb+Au at SPS-energies (preliminary data). As in case of S+Au, no significant excess (beyond systematic uncertainties) compared to the expectation from hadronic decays was found

Bildquelle: Ceres-Coll., Recent results from Pb-Au collisions at 158 GeV/c per nucleon obtained with the CERES spectrometer, Proc. QM99 Torino, Fig. 8

((evtl. Bildvorlage via J.Stachel))

Fig. 11: Ceres and Persephone (Rom, 496 BC. Quelle ggf.: Uffizien, Florence). ************************************

See GSI-Nachrichten 3/1999 for shortened and printed version of the article.