J/psi production in proton-nucleus collisions at 158 and 400 GeV
R. Arnaldi, K. Banicz, J. Castor, B. Chaurand, W. Chen, C. Cicalo, A. Colla, P. Cortese, S. Damjanovic, A. David, A. de Falco, A. Devaux, L. Ducroux, H. En'yo, J. Fargeix, A. Ferretti, M. Floris, A. Foerster, P. Force, N. Guettet, A. Guichard, H. Gulkanian, J. M. Heuser, M. Keil, Z. Li, C. Lourenco, J. Lozano, F. Manso, P. Martins, A. Masoni, A. Neves, H. Ohnishi, C. Oppedisano, P. Parracho, P. Pillot, T. Poghosyan, G. Puddu, E. Radermacher, P. Ramalhete, P. Rosinsky, E. Scomparin, J. Seixas, S. Serci, R. Shahoyan, P. Sonderegger, H. J. Specht, R. Tieulent, A. Uras, G. Usai, R. Veenhof, H. K. Woehri
aa r X i v : . [ nu c l - e x ] N ov J/ ψ production in proton-nucleus collisionsat 158 and 400 GeV R. Arnaldi a , K. Banicz b,c , J. Castor d , B. Chaurand e , W. Chen f , C. Cical`o g ,A. Colla h , P. Cortese a , S. Damjanovic b,c , A. David b,i , A. de Falco j ,A. Devaux d , L. Ducroux k , H. En’yo l , J. Fargeix d , A. Ferretti h , M. Floris j ,A. F¨orster b , P. Force d , N. Guettet b,d , A. Guichard k , H. Gulkanian m ,J. M. Heuser l , M. Keil b,i , Z. Li f , C. Louren¸co b , J. Lozano i , F. Manso d ,P. Martins b,i , A. Masoni g , A. Neves i , H. Ohnishi l , C. Oppedisano a ,P. Parracho i , P. Pillot k , T. Poghosyan m , G. Puddu j , E. Radermacher b ,P. Ramalhete b,i , P. Rosinsky b , E. Scomparin a , J. Seixas i , S. Serci j ,R. Shahoyan b,i , P. Sonderegger i , H. J. Specht c , R. Tieulent k , A. Uras j ,G. Usai j , R. Veenhof i , H. K. W¨ohri i a INFN, Sezione di Torino, Italy b CERN, 1211 Geneva 23, Switzerland c Physikalisches Institut der Universit¨at Heidelberg, Germany d Universit´e Blaise Pascal and CNRS-IN2P3, Clermont-Ferrand, France e LLR, Ecole Polytechnique and CNRS-IN2P3, Palaiseau, France f BNL, Upton, NY, USA g INFN, Sezione di Cagliari, Italy h Dipartimento di Fisica Sperimentale dell’ Universit`a di Torino and INFN, Torino, Italy i Instituto Superior T´ecnico, Lisbon, Portugal j Dipartimento di Fisica dell’ Universit`a di Cagliari and INFN, Cagliari, Italy k IPNL, Universit´e Claude Bernard Lyon-I and CNRS-IN2P3, Villeurbanne, France l RIKEN, Wako, Saitama, Japan m YerPhI, Yerevan Physics Institute, Yerevan, Armenia
Abstract
The NA60 experiment has studied J/ ψ production in p-A collisions at 158and 400 GeV, at the CERN SPS. Nuclear effects on the J/ ψ yield havebeen estimated from the A-dependence of the production cross section ratios σ A J /ψ /σ Be J /ψ (A=Al, Cu, In, W, Pb, U). We observe a significant nuclear sup-pression of the J/ ψ yield per nucleon-nucleon collision, with a larger effectat lower incident energy, and we compare this result with previous obser-vations by other fixed-target experiments. An attempt to disentangle thedifferent contributions to the observed suppression has been carried out by Preprint submitted to Physics Letters B September 11, 2018 tudying the dependence of nuclear effects on x , the fraction of the nucleonmomentum carried by the interacting parton in the target nucleus. Keywords:
1. Introduction
The study of charmonium production in hadronic collisions is an interest-ing test of our understanding of the physics of strong interactions. While theproduction of the cc pair can be addressed in a perturbative-QCD approach,the subsequent binding of the pair is an essentially non-perturbative process,involving soft partons and occurring on a rather long timescale ( > ψ production in p-p collisions isstill missing [2]. In p-A interactions, the heavy-quark pair is created in thenuclear medium, and the study of its evolution towards a bound state canadd significant constraints to the models. For example, the strength of theinteraction between the evolving cc pair and the target nucleons, that canlead to a break-up of the pair and consequently to a suppression of the J/ ψ yield, may depend on its quantum state at the production level (color-octetor color-singlet), and on the kinematic variables of the pair [3, 4]. In addi-tion to final-state effects, also initial-state effects may influence the observedJ/ ψ yield in p-A. In particular, parton shadowing in the target nucleus [5]may suppress (or enhance, in the case of anti-shadowing) the probability ofproducing a J/ ψ , while the energy loss of the incident parton [6] in the nu-clear medium, prior to cc production, may significantly alter the J/ ψ crosssection and kinematic distributions. Furthermore, effects as final-state en-ergy loss and the presence of an intrinsic charm component in the protonmay also play a significant role [7]. Clearly, the correct understanding andthe disentangling of the various nucleus-related effects on J/ ψ production isa non-trivial task, which poses significant challenges to theory, but at thesame time offers important insights on the J/ ψ production and interactionmechanisms. Finally, a suppression of the J/ ψ has been proposed a longtime ago as a signature of the formation, in ultrarelativistic nucleus-nucleuscollisions, of a state where quarks and gluons are deconfined (Quark-GluonPlasma) [8]. Results from p-A collisions, taken in the same kinematic con-ditions of A-A, and properly extrapolated to nucleus-nucleus collisions, aretherefore necessary to calibrate the contribution of the various cold nuclear2atter effects to the overall observed suppression [9, 10].Having to deal with a rather complicated interplay of various physicalprocesses, the availability of accurate sets of data, spanning large intervals inthe incident proton energy, and covering large x F and p T regions, is essentialfor a thorough understanding of the involved mechanisms. At fixed targetenergies, high-statistics J/ ψ samples have been collected in recent years bythe DESY experiment HERA-B [11], at 920 GeV incident energy, by E866 [12]at FNAL at 800 GeV, and by the CERN-SPS experiment NA50 at 400 and450 GeV [13].In this Letter, we present an extension of these measurements towardslower energies, carried out by the NA60 experiment [14]. J/ ψ productionhas been studied in p-A collisions at 158 GeV, the same energy used inA-A collisions at the CERN SPS. Data have also been taken with a 400GeV beam, in order to provide a result that can be compared with previoussets of p-A data taken by the NA50 experiment [13]. We have performed asystematic study of nuclear effects by analyzing the A − dependence of theJ/ ψ production cross section on seven different target nuclei (Be, Al, Cu, In,W, Pb and U). The results are relative to a region close to midrapidity, andare presented differentially in x F and x and compared with previous resultsfrom the higher- √ s measurements mentioned above. The influence of partonshadowing in the nuclear targets on the results is also discussed, based onvarious recent parametrizations of this effect [5, 15, 16, 17]. A comparisonbetween our results at the two energies as a function of x , the fraction of thenucleon momentum carried by the parton in the nuclear target that producesthe J/ ψ , is carried out. By doing so, the shadowing effects can be factorizedand the importance of the other nuclear effects can be investigated.
2. Data analysis
The NA60 experiment has measured muon pair production in p-A andA-A collisions at the CERN SPS. Its experimental apparatus was based ona muon spectrometer (MS), positioned downstream of a hadron absorberwith a total thickness of 12 nuclear interaction lengths ( λ I ). The MS iscoupled to a vertex spectrometer (VT) based on Si pixel detectors. Theexperiment triggered on muon pairs detected in the MS, which were thenmatched, during reconstruction, to the corresponding tracks in the VT. Fordetails on the detector set-up and matching between the MS and VT we referto [14]. 3he J/ ψ mesons have been identified through their decay to a muon pair.In the p-A data taking, the SPS proton beam, with an average intensity of5 · s − , was hitting a target system composed of nine sub-targets withthicknesses between 0.005 and 0.012 λ I , with relative spacing between tar-gets from 0.8 to 1 cm. To limit the possible influence of position-dependentsystematic effects on the measurement of the nuclear dependence of the J/ ψ yield, the subtargets were placed in a mixed-A order, namely Al, U, W, Cu,In, Be, Be, Be and Pb.The analysis described in this Letter has been carried out on a data sampleconsisting of 2.8 · events at 158 GeV, and 1.5 · events at 400 GeV,containing a dimuon reconstructed in the MS. The mass region m µµ > , dominated by J/ ψ decays, contains 3.2 · and 2.1 · events at158 and 400 GeV, respectively. For about 50% of these events, each of thetwo MS tracks can be matched in direction and momentum with a track inthe VT. Such a value of the dimuon matching efficiency is due the smallersolid angle coverage of the VT with respect to the MS and to the inefficiencyof the pixel sensors as will be discussed below. A Monte-Carlo simulationhas shown that the contamination from events where at least one of the MStracks is matched to a wrong VT track is negligible in the J/ ψ mass region.For events containing a pair of hard muons from the decay of a J/ ψ (theaverage muon momentum is 27 GeV/c), the point of closest approach ofthe two muons gives an accurate estimate of their production point, with aresolution of ∼ µ m. In Fig. 1 we show the distribution of the longitudinalcoordinates of the dimuon production vertex, for m µµ > . Thepeaks corresponding to the position of the various production targets areclearly visible and well separated, showing that the nuclear target where theJ/ ψ has been produced is unambiguously determined.The MS detects J/ ψ produced in the rapidity range 3 < y lab <
4. However,the acceptance of the VT for matched tracks from the J/ ψ decay is target-dependent, the more downstream targets covering smaller rapidities. Wehave therefore selected events produced in the range 3.2 < y lab < < cos θ CS < ψ measurement extends down to zero p T . Theacceptance varies by about 30% in the p T range accessible with the collectedstatistics ( p T . ψ signal has been extracted for each target nucleus at the two4 vertex (cm)-9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 E ve n t s Al U W Cu In Be1 Be2 Be3 Pb
Figure 1: The distribution of the longitudinal coordinate of the point of closest approachof the opposite sign muon pair, for m µµ > . energies by fitting the opposite sign mass spectra, in the region m µµ > , with a superposition of the mass shapes of the expected dimuonsources. These include the Drell-Yan process (DY), the semi-leptonic decaysof correlated D-meson pairs ( DD ), the J/ ψ and the ψ ′ resonances. Theexpected mass distributions for the DY and DD contributions have beencalculated with PYTHIA [18], using the GRV94LO [19] parton distributionfunctions. The J/ ψ events have been generated, for the 400 GeV data, usingthe y and p T distributions measured with good accuracy by NA50 [20]. At158 GeV, the differential distributions have been tuned directly on the data,using an iterative procedure. The events have been tracked through the set-up and then reconstructed with the same algorithm used for real data. Thecontribution from the combinatorial pair background due to π and K decays(completely negligible in the J/ ψ mass region) has been estimated through anevent mixing technique [14]. In Fig. 2 we show, as an example, the opposite-sign dimuon invariant mass distribution, relative to the kinematical domainspecified above, for p-In interactions at 158 GeV. The quality of all the fitsto the invariant mass spectra is satisfactory (with χ /ndf ranging from 0.7 to1.5). The mass resolution at the J/ ψ peak is ∼
70 MeV/c , and the number5f J/ ψ events ranges from ∼
800 to ∼ ψ peak is very small ( < / ndf (cid:1) mm (cid:1)(cid:1) (GeV/c )1.5 2 2.5 3 3.5 4 4.5 5 E ve n t s / . G e V / c / ndf (cid:2) p-In Figure 2: Fit to the µ + µ − invariant mass spectrum for p-In collisions at 158 GeV. Thedashed line represents the Drell-Yan process, the dotted lines the charmonium resonances,the dashed-dotted line the DD contribution, the thin continuous line the combinatorialbackground. The thick continuous line is the sum of all the contributions. Nuclear effects have been parametrized by fitting the A − dependence ofthe production cross section with the simple power law σ pA J /ψ = σ pp J /ψ · A α , andthen studying the evolution of α as a function of various kinematic variables.Alternatively, nuclear effects have been quantified by fitting the data in theframework of the Glauber model, having as input parameters the densitydistributions for the various nuclei. The model gives as output the so-calledJ/ ψ absorption cross section σ abs J /ψ . Clearly, both α and σ abs J /ψ represent effectivequantities, including the contribution of the various sources of nuclear effectsdetailed in the Introduction.The nuclear effects on J/ ψ production have been evaluated starting fromthe cross section measured for each target, normalized to the cross section6or the lightest one (Be): σ J /ψA . Aσ J /ψBe . A Be = N J /ψA . AN inc A · N targ A · A A · ǫ A , N J /ψBe . A Be N inc Be · N targ Be · A Be · ǫ Be (1)where, for the target with mass number A , N J /ψA is the number of J/ ψ events, N inc A is the number of incident protons, N targ A is the number of target nucleiper unit surface, A A is the J/ ψ acceptance and ǫ A is the detection efficiency.When building these ratios, the results obtained with the three Be targetshave been averaged. The use of relative cross sections brings several advan-tages. In particular, the number of incident protons cancels out (a smallattenuation factor, due to the inelastic scattering of beam particles, whichis 6% for the most downstream target, has been corrected for) since all thetargets were simultaneously exposed to the beam, which had a transversedimension much smaller than that of the targets. Furthermore, the fractionof the detection efficiency related to the MS also cancels out. In fact, thisdetector cannot distinguish the target where the dimuon has been produced.This is due to the presence of the thick hadron absorber, to the large dis-tance between the target system and the MS tracking chambers (6 to 16 m),and to the closely spaced targets. Therefore, the muon detection efficiencyis independent of the production target.Contrary to the MS case, the angular acceptance covered by the VT isslightly target dependent. Therefore, non-uniformities in the efficiency of acertain pixel plane lead to a rapidity-dependent inefficiency which is differentfrom target to target. The efficiency of each of the 15 VT planes has beenestimated using a modified track reconstruction algorithm that excludes theplane under study. For each reconstructed track we then check the presence ofa hit in that plane in a fiducial region around the intersection of the track withthe plane. Efficiency values are calculated on a run-per-run basis ( . ∼ , if the trackstatistics in the sub-region under study is large enough (more than 40 tracks).Otherwise, contiguous regions are grouped in order to reach a statisticallysignificant track sample. The distribution of the efficiency values for thevarious regions of the VT is peaked at ∼ ψ acceptances, estimatedby means of the Monte-Carlo simulation, range between 14.9% and 23.0%in the kinematic range 3.2 < y lab < | cos θ CS | < A / L Be ( A A · ǫ A ) / ( A Be · ǫ Be )(158) ( A A · ǫ A ) / ( A Be · ǫ Be )(400)Al 1.048 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Table 1: The relative values, target by target, of the integrated luminosities, and of theproducts A A · ǫ A . We have chosen as reference value the most upstream Be target. TheAl target was not in place during the data taking at 400 GeV. range corresponds to the center-of-mass rapidity windows 0.28 < y < − . < y < ψ acceptance times detection efficien-cies. For completeness, we also report the relative integrated luminosities L A / L Be = ( N inc A · N targ A ) / ( N inc Be · N targ Be ).
3. Results
In Fig. 3 we present the cross section ratios ( σ J /ψi /A i ) / ( σ J /ψBe /A Be ), at158 and 400 GeV, where A i is the nuclear mass number of target i . Theresults are shown as a function of L , the mean thickness of nuclear mat-ter crossed by the J/ ψ in its way through the target nucleus. The L val-ues have been computed with the Glauber model, using realistic densitydistributions for the various nuclei [21]. The quoted systematic uncertain-ties include contributions, quadratically combined, coming from the uncer-tainty on i) the measurement of the target thicknesses ( ≤ ψ acceptance, due to the choice of the rapidity distribution adopted inthe Monte-Carlo calculation ( ≤ ≤ ±
10% the estimated efficiency values of the VT pixel detectors. We onlyquote, for each incident energy, the fraction of the systematic uncertainty8hich is not common to all the points, the only one relevant when plot-ting relative cross sections. Fig. 3 shows, for both datasets, a suppressionof the J/ ψ yield when moving from light to heavy targets, and, in particu-lar, a larger suppression for the 158 GeV data sample. Using the Glaubermodel, we have estimated σ abs J /ψ (158 GeV) = 7 . ± . ± . σ abs J /ψ (400 GeV) = 4 . ± . ± . α J /ψ (158 GeV) = 0 . ± . ± . α J /ψ (400 GeV) = 0 . ± . ± . σ abs J /ψ = 4 . ± . ψ yield. L (fm)0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 ) B e / A B e y J / s ) / ( i / A i y J / s ( Statistical errorsSystematic errors
Figure 3: Cross sections for J/ ψ production in p-A collisions, normalized to the p-Be J/ ψ cross section. The squares represent the 158 GeV results, the circles refer to 400 GeV.The lines are the fit results. To get further insight on the dependence of nuclear effects on the J/ ψ kinematic variables we compare in Fig. 4, as a function of x F , the α valuesobtained in this analysis with those from previous experiments in the fixed-target energy range. For this purpose, the NA60 sample has been subdivided9nergy (GeV) x F α
158 0 . ÷ .
15 0 . ± . ± . . ÷ .
20 0 . ± . ± . . ÷ .
25 0 . ± . ± . . ÷ .
30 0 . ± . ± . . ÷ .
40 0 . ± . ± . − . ÷ − .
025 0 . ± . ± . − . ÷ .
025 0 . ± . ± . . ÷ .
075 0 . ± . ± . . ÷ .
125 0 . ± . ± . Table 2: α as a function of x F for the 158 and 400 GeV data sample. The first quoteduncertainty is statistical, the second is the systematic one. into 5 x F bins at 158 GeV (4 at 400 GeV), covering the region 0 . < x F < .
40 ( − . < x F < . α values.Two main features emerge from this comparison. First, when going fromnegative towards positive x F , α steadily decreases. This effect was alreadyknown from the data of HERA-B and E866, taken at rather similar incidentproton energies (920, 800 GeV), and a similar effect might be present in ourresults at 158 GeV, even if the size of the errors is not negligible. Second,at a constant x F , the α values are lower when the incident proton energy issmaller, as can be seen when comparing the HERA-B/E866 results with ourresults at 400 and 158 GeV. On the other hand, it is also worth mentioningthat from NA3 results on J/ ψ production at 200 GeV [22] one extracts α values which are in partial contradiction with these observations, being simi-lar to those obtained with the higher energy data samples (HERA-B/E866).We also note that in our calculation of the NA3 α values, performed startingfrom their measured J/ ψ cross section ratios between p-Pt and p-p collisions,the small bias [23] induced by the use of a light target in the determinationof α has been corrected for.A satisfactory theoretical interpretation of the complex observed patternis missing for the moment. Various works have underlined the importance ofseveral effects, including final state break-up, parton shadowing, initial andfinal state energy loss, and the presence of a charm component in the nucleonwavefunction [7, 24, 25]. However, the relative weight of these effects is still10nder debate. F x-0.4 -0.2 0 0.2 0.4 0.6 0.8 a HERAB 920 GeVE866 800 GeVNA50 450 GeVNA60 400 GeVNA3 200 GeVNA60 158 GeV
Figure 4: The x F dependence of the α parameter. Open triangles correspond to HERA-Bresults, open squares to E866, open circles to NA50 (450 GeV), closed circles to NA60 (400GeV), closed squares to NA60 (158 GeV), open stars to NA3. The error bars representthe quadratic sum of statistical and systematic uncertainties. The new results from NA60 at 400 and 158 GeV have been obtained withthe same experimental apparatus and in very similar running conditions.Their direct comparison may therefore be a clean testing ground for models.Interesting information can also be obtained by a simple comparison of nu-clear effects as a function of various kinematic variables. As an example, wenow consider the x dependence of the α parameter, x = m T / √ s · exp( − y )being the fraction of the momentum of the target nucleon carried by theparton which produces the J/ ψ . This kinematic variable is particularly Such a relation, commonly adopted, implies that the 2 → ψ production process gg → J /ψg is, effectively, that of a 2 → gg → J /ψ ) i.e. thefinal state gluon is very soft. The alternative 2 → p T ≪ p L in the c.m. frame. Due to thelow √ s , this is only approximately true and induces a smearing in the estimation of x .However, in the investigated kinematic domain, the effect is similar for the two energies.The resulting ∼
10% shift in the x values does not affect the discussion in the text. x is the quantity that determines the amount of shad-owing in the target nucleus (or anti-shadowing in our kinematic range).Furthermore, the center of mass energy of the J/ ψ -N system, which is arelevant quantity for the final state break-up of the J/ ψ in the cold nu-clear medium, is, to a very good approximation, a function of x alone( √ s J /ψ N ∼ m J /ψ (cid:14) p (1 + x )/ x ).Since our results at the two energies cover, with good approximation, thesame x domain, we have performed an analysis of nuclear effects in five x bins, covering the region 0.08 < x < α as a functionof x for the two energies under study. We also show in the same plot theexpected α value that would be obtained if shadowing were the only nucleareffect to be present. Various parameterizations of nuclear shadowing havebeen considered [5, 15, 16, 17], including the recent EPS09 result, where anestimate of the uncertainty of the calculation was also carried out. Lookingat Fig. 5 we see that shadowing alone would lead to α values larger than 1 inour x acceptance. This result implies that the other nuclear effects producea significant suppression of the J/ ψ yield.The other remarkable feature emerging from Fig. 5 is that α is not thesame at a fixed x , for the two values of √ s . In our analysis, the system-atic uncertainties related to the measurement of α are partly correlated,in particular those corresponding to the measurement of the J/ ψ detectionefficiency. Therefore, in Fig. 6 we plot, as a function of x , the quantity∆ α = α (400GeV) − α (158GeV) which is affected by a significantly smallersystematic uncertainty. The results clearly indicate that ∆ α = 0 (in the cov-ered x range we have h ∆ α i = 0 . ± . ± . α does not vary appreciably in the x region under study.Having factorized the effect of shadowing, which is the same at the twoenergies at fixed x , the observation of ∆ α > x , the kine-matics of the collision of the J/ ψ with the target nucleons is the same at158 and 400 GeV, the observed effect may be related to a change in the J/ ψ break-up cross section at fixed √ s J /ψ N . A different weight of the color-octetand color-singlet precursor c ¯ c states in the production process between thetwo energies could be a natural explanation. However, Non-Relativistic QCD(NRQCD) calculations carried out in the fixed target energy domain [27],only show a weak √ s dependence of the color-octet and color-singlet rela-tive contributions. Therefore, a strong variation of the J/ ψ break-up crosssection with √ s seems unlikely. In such a case one should observe, for the12wo incident proton energies, the same α at a given x in contrast with ourobservation. In this scenario, processes different from final state break-upand parton shadowing must be advocated in order to explain the observednuclear effects on J/ ψ production. In particular, effects such as the initialstate energy loss of the incident parton might play an important role anddeserve further investigation. x0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 a EPS09EPS08EKS98HKN400 GeV158 GeV
Figure 5: α as a function of x , for the 158 and 400 GeV data samples. Also shownin the plot is a calculation of α as resulting from shadowing alone. Various shadowingparameterizations have been considered.
4. Conclusions
The NA60 experiment has measured J/ ψ production in p-A collisions atthe CERN SPS, at 400 and 158 GeV, the latter energy being the lowest onewhere a detailed systematic study has been performed. The results showa suppression of the J/ ψ yield in cold nuclear matter, which is larger atlower incident energy. A comparison with results from previous experimentsindicates that nuclear effects on J/ ψ production, at constant x F , exhibita strong √ s -dependence. The observation of different α values at the twoenergies, for a constant x , may suggest a strong change in the J/ ψ break-upcross section and/or the presence of other effects like initial state energy loss.13 x0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 aD -0.04-0.0200.020.040.060.080.10.120.14 Figure 6: ∆ α = α (400GeV) − α (158GeV), as a function of x . The empty boxes around thepoints represent the uncorrelated systematic uncertainty. The systematic error commonto all the points is shown as a filled box around the ∆ α = 0 line. These data can also provide an important baseline for heavy-ion collisionstudies.
References [1] J.P. Lansberg, Eur. Phys. J.
C60 , 693(2009).[2] N. Brambilla et al.(Quarkonium Working Group), Eur. Phys. J.
C71 ,1534 (2011).[3] R. Vogt, Nucl. Phys.
A700 , 539 (2002).[4] B.Z. Kopeliovich and B.G. Zakharov, Phys. Rev.
D44 , 3466 (1991).[5] K.J. Eskola, H. Paukkunen and C.A. Salgado, JHEP , 065 (2009).[6] S. Gavin and J. Milana, Pys. Rev. Lett. , 1834 (1992).[7] R. Vogt, Phys. Rev. C61 , 035203 (2000).148] T. Matsui and H. Satz, Phys. Lett.
B178 , 416 (1986).[9] B. Alessandro et al.(NA50), Eur. Phys. J.
C39 , 335 (2005).[10] R. Arnaldi et al. (NA60), Phys. Rev. Lett. , 132302 (2007).[11] I. Abt et al. (HERA-B), Eur. Phys. J. C60 , 525 (2009).[12] M.J. Leitch et al. (E866), Phys. Rev. Lett. , 3256 (2000).[13] B. Alessandro et al. (NA50), Eur. Phys. J. C48 , 329 (2006).[14] R. Arnaldi et al. (NA60), Eur. Phys. J.
C59 , 607 (2009).[15] M. Hirai et al., Phys. Rev.
C70 C9
61 (1999).[17] K.J. Eskola et al., JHEP
102 (2008).[18] T. Sj¨ostrand et al., Comp. Phys. Comm. , 238 (2001).[19] M. Gl¨uck et al., Z. Phys.
C67 , 433 (1995).[20] B. Alessandro et al. (NA50), Eur. Phys. J.
C33 , 31 (2004).[21] H. DeVries, C.W. DeJager and C. DeVries, Atomic Data and Nucl. DataTables , 495 (1987).[22] J. Badier et al. (NA3), Z. Phys. C20 , 101 (1983).[23] R. Shahoyan, PhD thesis, Universidade T´ecnica de Lisboa, 2001.[24] V.-N. Tram and F. Arleo, Eur. Phys. J.
C61 , 847 (2009).[25] C. Lourenco et al., JHEP , 014 (2009).[26] E.G. Ferreiro et al., Phys.Lett.
B680
50 (2009).[27] R. Vogt, Phys. Rev.