Measurement of inclusive $J/\psi$ suppression in Au+Au collisions at $\sqrt{s_{NN}}$ = 200 GeV through the dimuon channel at STAR

$J/\psi$ suppression has long been considered a sensitive signature of the formation of the Quark-Gluon Plasma (QGP) in relativistic heavy-ion collisions. In this letter, we present the first measurement of inclusive $J/\psi$ production at mid-rapidity through the dimuon decay channel in Au+Au collisions at $\sqrt{s_{NN}}$ = 200 GeV with the STAR experiment. These measurements became possible after the installation of the Muon Telescope Detector was completed in 2014. The $J/\psi$ yields are measured in a wide transverse momentum ($p_{\rm{T}}$) range of 0.15 GeV/$c$ to 12 GeV/$c$ from central to peripheral collisions. They extend the kinematic reach of previous measurements at RHIC with improved precision. In the 0-10% most central collisions, the $J/\psi$ yield is suppressed by a factor of approximately 3 for $p_{\rm{T}}>5$ GeV/$c$ relative to that in p+p collisions scaled by the number of binary nucleon-nucleon collisions. The $J/\psi$ nuclear modification factor displays little dependence on $p_{\rm{T}}$ in all centrality bins. Model calculations can qualitatively describe the data, providing further evidence for the color-screening effect experienced by $J/\psi$ mesons in the QGP.

Abstract J/ψ suppression has long been considered a sensitive signature of the formation of the Quark-Gluon Plasma (QGP) in relativistic heavy-ion collisions.
In this letter, we present the first measurement of inclusive J/ψ production at mid-rapidity through the dimuon decay channel in Au+Au collisions at √ s NN = 200 GeV with the STAR experiment. These measurements became possible after the installation of the Muon Telescope Detector was completed in 2014. The J/ψ yields are measured in a wide transverse momentum (p T ) range of 0.15 GeV/c to 12 GeV/c from central to peripheral collisions. They extend the kinematic reach of previous measurements at RHIC with improved precision. In the 0-10% most central collisions, the J/ψ yield is suppressed by a factor of approximately 3 for p T > 5 GeV/c relative to that in p+p collisions scaled by the number of binary nucleon-nucleon collisions. The J/ψ nuclear modification factor displays little dependence on p T in all centrality bins. Model calculations can qualitatively describe the data, providing further evidence for the color-screening effect experienced by J/ψ mesons in the QGP.

Introduction
Among the primary goals of high-energy heavy-ion physics are the creation of the Quark-Gluon Plasma (QGP) and the study of its properties [1]. These studies are being carried out at the Relativistic Heavy Ion Collider (RHIC) and the Large Hadron Collider (LHC). Among the various probes of the QGP, quarkonia play a special role as they are expected to dissociate in the medium when the Debye radius, inversely proportional to the medium temperature, becomes smaller than their size [2]. Strong suppression of the J/ψ meson with respect to its yield in p+p collisions scaled by the number of binary nucleonnucleon collisions has been observed at high transverse momenta (p T ) in central heavy-ion collisions at both RHIC and LHC energies [3,4,5,6,7,8,9,10].
The level of suppression is beyond that expected from Cold Nuclear Matter (CNM) effects [11,12,13], which include modifications to the parton distribution function in nuclei [14,15], nuclear absorption [16], and radiative energy loss [17]. This suggests that the reduction of the high-p T J/ψ yield is, at least partially, due to the presence of the hot medium and the color-screening effect is believed to be the underlying mechanism. The real part of the cc potential can get color-screened statically in the medium [2], resulting in a broadening of the wave function, while the imaginary part of the potential is related to the dissociation of J/ψ arising from scattering with medium constituents. The latter is sometimes referred to as the dynamical color-screening effect or collisional dissociation [18,19,20]. Other effects have also been found to modify the observed J/ψ yield in heavy-ion collisions [21]. A prominent contribution arises from the regeneration of J/ψ from deconfined charm and anti-charm quarks in the medium. It is responsible for the reduced suppression of low-p T J/ψ's at the LHC compared to RHIC [7] due to the larger charm production cross-section at the former. Also, the pre-resonance cc pairs in color-octet states could undergo energy loss in the medium before quarkonia are formed [20]. Furthermore, significant feed-down contributions from excited charmonium states such as χ c and ψ(2S) (∼40% [22]) as well as from b-hadron decays (∼10-25% above 5 GeV/c [4]) add additional complications as the suppression level for mother particles in the medium could differ from that of directly produced J/ψ, i.e. ones not from decays. Model calculations, incorporating either continuous dissociation and regeneration throughout the medium evolution [23,24,25,26] or a complete melting of all J/ψ above the dissociation temperature and regeneration at the phase boundary [27,28] or collisional dissociation plus energy loss [20], can qualitatively describe the experimental measurements. To provide further constraints on models and ultimately help infer the medium temperature, detailed differential measurements of J/ψ suppression over a broad kinematic range with good precision are needed since the aforementioned effects depend on the momentum of the J/ψ as well as the collision geometry. Measurements through the dimuon decay channel are preferred compared to the dielectron channel because of the greatly reduced multiple scattering in the material and negligible bremsstrahlung.
In this letter, we present a new measurement of J/ψ suppression at midrapidity in Au+Au collisions at √ s NN = 200 GeV through the dimuon decay channel by the Solenoidal Tracker At RHIC (STAR) experiment [29]. The inclusive J/ψ sample used in this analysis includes decays from excited charmonia and b-hadrons. This measurement is made possible by the Muon Telescope Detector (MTD) designed for triggering on and identifying muons [30], which was completed in early 2014. Compared to previous mid-rapidity measurements through the dielectron channel at RHIC [3,4,5,6], the new results extend the kinematic reach towards high p T with better precision.

Experiment, dataset and analysis
The data sample used in this analysis was collected from Au+Au collisions at √ s NN = 200 GeV in 2014. Events were selected by a dedicated dimuon trigger, which requires at least two muon signals accepted by the MTD in coincidence with signals in the Zero Degree Calorimeters (ZDCs) [31]. The MTD consists of 122 modules made from multi-gap resistive plate chambers, providing timing information for particles passing through. It resides outside of the solenoid magnet at a radius of 403 cm, and covers about 45% in azimuth (ϕ) within the pseudo-rapidity range of |η| < 0.5. The magnet also acts as a hadron absorber amounting to 5 interaction lengths. Using double-ended readout strips, the timing resolution of the MTD is about 100 ps, and the intrinsic spatial resolutions are 1.4 cm and 0.9 cm in rϕ and beam (z) directions, respectively [32]. Variable numbers of MTD modules, ranging from 2 to 5 and located at the same η, are grouped into 28 trigger patches. The earliest signal in each trigger patch is picked up and accepted by the trigger system if its flight time (∆t trig ) falls into a pre-defined online trigger time window. The ∆t trig is the difference in time measured by the MTD and the start time provided by the Vertex Position Detector (VPD), which is a fast detector covering 4.24 < |η| < 5.1 [33]. In total, an integrated luminosity of 14.2 nb −1 was sampled by the dimuon trigger.
The main tracking device is the Time Projection Chamber (TPC) [34] immersed in a solenoidal magnetic field of 0.5 T and covering full azimuth within |η| < 1.0. The primary event vertex is reconstructed using TPC tracks, and required to be within ±100 cm to the center of STAR along the beam line and within 1.8 cm in radial direction. To reject pileup events, the vertex positions determined by the TPC and the VPD are required to agree within 3 cm along the beam direction. The collision centrality is determined by matching the multiplicity distribution of charged tracks from data to the Monte Carlo Glauber model [35]. The selected charged tracks are within |η| < 0.5 and have Distances of Closest Approach (DCA) to the primary vertex of less than 3 cm.

Muon identification and J/ψ signal
Since particles of low momenta are mostly absorbed in the material in front of the MTD, only charged tracks with p T > 1.3 GeV/c are accepted. To assure high quality, the number of TPC space points used for track reconstruction is required to be no less than 15. The ratio of the number of used to the maximum possible number of TPC space points is required to be larger than 0.52 in order to reject split tracks. Furthermore, a track's DCA to the primary vertex needs to be smaller than 1 cm to get accepted. It is then refit including the primary vertex to improve the momentum resolution. To identify muon candidates, the specific energy loss (dE/dx) measured in the TPC, quantified as nσ π , is used: Here (dE/dx) measured is the measured energy loss in the TPC, (dE/dx) π theory is the expected energy loss for a pion based on the Bichsel formalism [36] and σ(ln(dE/dx)) stands for the resolution of the ln(dE/dx) measurement. Since muons lose more energy per unit of path length by about half of the dE/dx resolution than pions, an asymmetric cut of −1 < nσ π < 3 is used.
To take advantage of the MTD, tracks are propagated from the outermost  Table 1.
Detector used Muon PID cuts

Efficiency correction
Corrections for signal reconstruction efficiency and detector acceptance are evaluated using a combination of detector simulation and data-driven methods. They include efficiencies for the TPC tracking, MTD matching, particle identification (PID) of muons, and MTD triggering.
The TPC tracking efficiency, including the TPC acceptance, is evaluated by embedding simulated J/ψ signals into real events. The input J/ψ's, weighted with previously published p T distributions [3,4], are forced to decay into two muons and then passed through the GEANT3 detector simulations. The sim-ulation signals are digitized and embedded into real data, and the same reconstruction procedure as for the real data is applied. Since the TPC tracking efficiency depends strongly on the occupancy, the number of embedded J/ψ is set to be 5% of the event multiplicity to avoid any significant distortion to the TPC performance. Additional correction factors are applied to account for the different vertex distributions between data and embedding samples, as well as additional luminosity and centrality dependences of the TPC inefficiency in local areas which are not accounted for in the embedding. The efficiencies related to muon identification cuts on nσ π , ∆y, ∆z are extracted from embedding, while the ∆t tof efficiency is evaluated with the "tagand-probe" method using real data since the timing information is not simulated. In this approach, a "tag" muon and a "probe" muon are paired to construct the J/ψ signal. The tag muons are always selected with strict PID cuts in order to increase the signal-to-background ratio, while the probe muons are selected with the standard nσ π , ∆y, ∆z cuts as well as two cases of the ∆t tof cut, i.e. no ∆t tof cut and ∆t tof < 0.75 ns. The ratio of the J/ψ yields from the two cases as a function of the probe muon p T is parametrized as the ∆t tof cut efficiency for muons. Figure 2, lower panel, shows the nσ π , ∆y plus ∆z, and ∆t tof cut efficiencies as well as the combined muon PID efficiency as a function of muon p T . The discontinuity at 3 GeV/c is due to the change in the ∆y and ∆z cuts (see Table 1). The muon PID efficiency is about 73% at p T = 1.3 GeV/c, and reaches a plateau of about 85% at high p T .  In order to extract the total J/ψ reconstruction efficiency, single muon efficiencies determined from data are applied to the J/ψ simulation, which takes the decay kinematics properly into account using the PYTHIA event generator [40].

Systematic uncertainties
Signal extraction. Variations are made to different aspects of the signal extraction procedure and the maximum differences from the default values are taken as the systematic uncertainties. When obtaining the normalization factors for the mixed-event background, the fit range is varied and the fit function is changed from first-order to zeroth-order polynomial. To extract the raw J/ψ counts, the binning of the invariant mass distributions is changed, and so is the fit range.
Different functional forms, such as a Crystal-ball function [41] and line-shapes from tuned simulation, are used for signal shape, while polynomial functions of different orders are substituted for describing the residual background. Finally, the bin-counting method, with the residual background contribution removed, is tried.
TPC tracking. The uncertainties in the TPC tracking efficiency are evaluated by changing the track quality cuts simultaneously in the data analysis and in extracting the tracking efficiency from the embedding sample, and repeating the whole procedure to obtain the corrected J/ψ yields. The maximum differences from the default case are seen to be almost independent of J/ψ p T for 0-80% centrality bin. A constant fit gives a p T -independent uncertainty of 5.8%. For finer centrality bins the same uncertainty is used, which covers most of the variation seen in these centrality bins. Furthermore, an overall 2% uncertainty is assigned for the correction factor used to account for the mismatch of the vertex distributions between data and embedding. An uncertainty of 0.2%-6.1% from central to peripheral events is associated with the correction of the luminosity and centrality dependent TPC inefficiencies. An additional 5% overall uncertainty is assigned based on the comparison of the like-sign muon pair yields in different luminosity profiles. ing response efficiency as a function of J/ψ p T . Secondly, the uncertainty from the assumption of using the efficiency template for top modules is estimated by taking the average absolute difference between the response efficiency curves of bottom modules and the template efficiency. Furthermore, the MTD matching efficiencies extracted from simulation and from cosmic ray data are compared and the difference is taken as an additional source of uncertainty. The total uncertainty on the MTD matching efficiency is taken as the quadratic sum of these three sources. It is 9.1% for 0.15 < p T < 1 GeV/c and decreases to 1.0% at the highest p T bin.
Muon PID. The uncertainties in the nσ π , ∆y, ∆z cut efficiencies, extracted from the embedding sample, are estimated the same way as the TPC tracking efficiency. For the ∆t tof cut efficiency, the uncertainty comes mainly from the statistical errors on the data points used to extract this efficiency. It is evaluated by randomly changing the data points independently within their individual errors, fitting the randomized data points, and taking the root-mean-square of the resulting efficiency distributions in each muon p T bin. The total uncertainty on the muon PID efficiency, shown as the band around the efficiency curve in the lower panel of Fig. 2, is the quadratic sum of the two contributions. All the aforementioned uncertainties are listed in Table 2 for two representative J/ψ p T bins, i.e. 0.15 < p T < 1 GeV/c and 5 < p T < 6 GeV/c in the 0-80% centrality class. The total uncertainties are the quadratic sum of all the individual sources. The uncertainties are fully or largely correlated across different p T and centrality bins except for the signal extraction uncertainty.

Results and Discussion
The invariant yields of inclusive J/ψ within |y| < 0.5 as a function of p T , measured through the dimuon channel, are shown in Fig. 3  cation factor (R AA ): where N coll is the average number of binary nucleon-nucleon collisions in a given centrality bin and (  show the global uncertainties, which for this analysis include the 10% global uncertainty on p+p reference and the N coll uncertainties. Other measurements [3,4,6,8,9] and model calculations [23,24,25,20] are shown for comparison. In the 0-80% panel, the boxes at unity from left to right correspond to CMS, ALICE and STAR results, while for other panels the left band is for PHENIX and the right one for STAR. which is likely due to the combination of the CNM effects [11] and the dissociation in the QGP. In the 60-80% centrality bin, the normalization uncertainty is large. Within the current uncertainties, the J/ψ R AA shows little dependence on p T . A sizable suppression in the J/ψ yield is present up to the largest measured p T bin in central and semi-central collisions. There are several effects that could influence the p T dependence of R AA . The CNM effects decrease with increasing p T . High p T J/ψ's spend less time in the medium and are therefore less likely to be dissociated [45,46]. Furthermore, the relative contributions from b-hadron decays, whose suppression level is expected to be smaller than that of direct J/ψ [25], rise with increasing p T . Also shown in Fig. 4  tion are expected to be minimal, the J/ψ production in 0-10% central collisions is suppressed by a factor of 3.1 with a significance of 8.5σ, providing strong evidence for the color-screening effect in the deconfined medium. Also shown in Fig. 5 is the J/ψ R AA as a function of N part measured for Pb+Pb collisions at √ s NN = 2.76 TeV [8,9]. Here, the low-p T J/ψ's are above 0 GeV/c and high-p T J/ψ's are above 6.5 GeV/c. Inclusion of very low-p T J/ψ from coherent photoproduction in Pb+Pb collisions has negligible impact on the measured R AA values for N part > 50 [48]. The low-p T J/ψ's are much more suppressed in central and semi-central collisions at RHIC than at the LHC, likely due to the smaller charm quark production cross-section and thus smaller regeneration contribution at RHIC. On the other hand, the high-p T J/ψ R AA is systematically higher at RHIC for semi-central bins. This could be because the temperature of the medium created at the LHC is higher than that at RHIC, leading to a higher dissociation rate. Transport model calculations are consistent with the data at low p T , while the data lay mostly between the two model calculations  Fig. 5, also describes the data reasonably well in non-peripheral events [28]. In the SHM model, the charm quark production cross-section from the fixed-order plus next-to-leading logs calculations [49] is used as input. However, feed-down contributions from b-hadron decays have not been included.

Summary
In summary, we report the first measurements of inclusive J/ψ R AA through the dimuon decay channel at mid-rapidity in Au+Au collisions at √ s NN = 200 GeV by the STAR experiment at RHIC. Compared to previous dielectron measurements, the new results provide an improved measure of J/ψ suppression in the QGP with better precision and in a wider kinematic range. At low p T , the interplay of the CNM effects, dissociation, and regeneration results in an increasing suppression of J/ψ from peripheral to central collisions. At p T above 5 GeV/c, the J/ψ yield is significantly suppressed in central collisions, which is caused mainly by color-screening in the medium due to the presence of the QGP. While both the Tsinghua and TAMU transport models describe the centrality dependence of J/ψ R AA at low p T , their agreement with data degrades at high p T .
The new results presented in this letter will help constrain model calculations and deepen our understanding of the QGP properties.