Medium-induced modification of Z-tagged charged particle yields in Pb+Pb collisions at 5.02 TeV with the ATLAS detector

The yield of charged particles opposite to a Z boson with large transverse momentum ( p T ) is measured in 260 pb − 1 of pp and 1 . 7 nb − 1 of Pb þ Pb collision data at 5.02 TeV per nucleon pair recorded with the ATLAS detector at the Large Hadron Collider. The Z boson tag is used to select hard-scattered partons with specific kinematics, and to observe how their showers are modified as they propagate through the quark-gluon plasma created in Pb þ Pb collisions. Compared with pp collisions, charged-particle yields in Pb þ Pb collisions show significant modifications as a function of charged-particle p T in a way that depends on event centrality and Z boson p T . The data are compared with a variety of theoretical calculations and provide new information about the medium-induced energy loss of partons in a p T regime difficult to measure through other channels.


Medium-Induced Modification of Z-Tagged
The yield of charged particles opposite to a Z boson with large transverse momentum (p T ) is measured in 260 pb −1 of pp and 1.7 nb −1 of Pb þ Pb collision data at 5.02 TeV per nucleon pair recorded with the ATLAS detector at the Large Hadron Collider. The Z boson tag is used to select hard-scattered partons with specific kinematics, and to observe how their showers are modified as they propagate through the quarkgluon plasma created in Pb þ Pb collisions. Compared with pp collisions, charged-particle yields in Pb þ Pb collisions show significant modifications as a function of charged-particle p T in a way that depends on event centrality and Z boson p T . The data are compared with a variety of theoretical calculations and provide new information about the medium-induced energy loss of partons in a p T regime difficult to measure through other channels. DOI: 10.1103/PhysRevLett.126.072301 Collisions of heavy nuclei at ultrarelativistic energies at the Large Hadron Collider (LHC) and the Relativistic Heavy Ion Collider (RHIC) are understood to produce an extended region of hot and dense matter where partons exist in a deconfined state known as the quark-gluon plasma (QGP). The high density of unscreened color charges in the QGP causes the showers of hard-scattered partons with large transverse momentum (p T ) to be modified as they traverse the medium [1]. These modifications are observed in measurements of dijet and photonjet momentum imbalance [2][3][4][5], and in jet fragmentation functions [6,7].
The large integrated luminosity of Pb þ Pb collisions delivered during LHC Run 2 has enabled measurements of jets produced in association with a high-p T Z boson. At leading order, the Z boson and the jet are produced back to back in the azimuthal plane, with equal p T . Since Z bosons and their decay leptons, or similarly, photons, do not participate in the strong interaction and are not modified by the QGP [8,9], they provide an estimate of the p T and azimuthal direction of the partner hard-scattered parton before the developing shower is modified through interactions with the QGP [10,11]. Measurements of photon-tagged fragmentation functions at the LHC [12,13] and photon-hadron correlations at RHIC [14,15] used this feature to perform detailed studies of jet quenching. At fixed p T , jets balancing Z bosons and photons arise from processes with different Q 2 , and can test the sensitivity of the energy loss process to parton virtuality. Additionally, the use of isolated photons at low photon p T (≲60 GeV) is difficult due to the large hadrondecay background, motivating the use of Z bosons. A measurement of Z þ jet production with p Z T > 60 GeV by CMS demonstrates that the total p T carried inside the jet cone is decreased in Pb þ Pb events compared with that in pp events [16]. However, the modification of the jet's constituent particle p T distributions, or any lower p Z T selections, have not yet been studied.
This Letter presents a measurement of the yield of charged particles produced opposite in azimuth to a Z boson with p Z T > 15 GeV in Pb þ Pb and pp collisions at a nucleon-nucleon center-of-mass energy ffiffiffiffiffiffiffi ffi s NN p ¼ 5.02 TeV with the ATLAS detector at the LHC. The Pb þ Pb and pp data were recorded in 2018 and 2017, respectively, and correspond to integrated luminosities of up to 1.7 nb −1 and 260 pb −1 . The charged particles are required to have p ch T > 1 GeV and be approximately back to back with the Z boson in the transverse plane, with azimuthal separation Δϕ larger than 3π=4 [17]. In simulations of pp collisions, particles meeting these criteria reside primarily in the leading jet azimuthally opposite to the Z boson. The per-Z yields of charged particles, N ch , are reported as a function of p ch T , ð1=N Z Þðd 2 N ch =dp ch T dΔϕÞ, in pp and Pb þ Pb collisions. To quantify the modification resulting from the partons' propagation through the QGP, the ratio of particle yields between Pb þ Pb and pp collisions, I AA , is reported and compared with the expectations from theoretical calculations. This measurement explores phenomena similar to those in measurements of the photon-tagged jet fragmentation function [12]. However, requiring a reconstructed jet may result in a bias towards events with less energy loss than average [18][19][20]. Since there is no such requirement in this measurement, it provides additional insight into energy loss in an unbiased way, at low p Z=γ T values which have not yet been measured at the LHC and where theoretical models have not been tested.
The ATLAS experiment [21] is a multipurpose particle detector with a forward-backward symmetric cylindrical geometry and a near 4π coverage in solid angle. It consists of an inner tracking detector surrounded by a superconducting solenoid providing a 2 T axial magnetic field, electromagnetic and hadron calorimeters, and a muon spectrometer. The inner tracking detector covers the pseudorapidity range jηj < 2.5. It consists of silicon pixel, silicon microstrip, and transition radiation tracking detectors [22,23]. Lead/liquid-argon (LAr) sampling calorimeters provide electromagnetic (EM) energy measurements with high granularity. A steel/scintillator-tile hadron calorimeter covers the central pseudorapidity range (jηj < 1.7). Liquid-argon calorimeters with separate EM and hadronic compartments instrument the end cap (up to jηj ¼ 3.2) and forward (FCal, up to jηj ¼ 4.9) regions. The muon spectrometer surrounds the calorimeters and includes three air-core toroidal superconducting magnets with field integrals ranging between 2.0 and 6.0 T m, a system of precision tracking chambers, and fast detectors for triggering. During Pb þ Pb data taking, the muon system was operational for only 1.4 nb −1 of the total integrated luminosity. Thus the dimuon channel is analyzed only in this subset of data.
Events with a high-p T electron or muon are initially selected for analysis by the single-lepton triggers described in Refs. [24,25]. The centrality of Pb þ Pb events is defined using the total transverse energy measured in the FCal [4,26], ΣE Pb T . Pb þ Pb events are divided into three categories which correspond to the 0%-10%, 10%-30%, and 30%-80% centrality intervals in minimum-bias (MB) events, the smaller values indicating larger nuclear overlap regions and thus larger, hotter QGP regions. The orientation of the underlying event (UE) elliptic flow is determined from the azimuthal distribution of the FCal energy [27,28]. In pp events, the average number of interactions per bunch crossing ranged from 2 to 4, and thus all chargedparticle tracks are required to originate from the primary reconstructed vertex [29].
Monte Carlo simulations of ffiffi ffi s p ¼ 5.02 TeV pp collisions with Z bosons decaying in the dielectron and dimuon channels, as well as data-driven studies, are used to correct the data for bin migration and reconstruction inefficiencies. Generated events were passed through a GEANT4 simulation [30,31] of the ATLAS detector under the same conditions present during data taking and were digitized and reconstructed in the same way as the data. The Z boson events were generated at next-to-leading order (NLO) with the POWHEG-BOX v2 program [32][33][34][35]  Four million events were generated to serve as the simulation sample for pp collisions. To model Pb þ Pb events, fifteen million simulated pp events were overlaid at the detector-hit level with MB Pb þ Pb events in data. This data-overlay sample was reweighted on an event-by-event basis to match the ΣE Pb T distribution for Pb þ Pb events containing Z bosons.
The Z bosons in pp and Pb þ Pb events are reconstructed in opposite-sign dielectron and dimuon decay channels using procedures similar to those described in Refs. [9,40]. Reconstructed electrons are required to have a transverse momentum p e T > 20 GeV, to lie within the fiducial acceptance of the EM barrel (jη e j < 1.37) or end cap (1.52 < jη e j < 2.47) detectors, and to satisfy "loose" likelihood-based identification criteria, which have been optimized separately for pp and Pb þ Pb events [41]. Reconstructed muons are required to have a transverse momentum p μ T > 20 GeV, to lie within the fiducial acceptance of the muon spectrometer (jη μ j < 2.5), and to pass the "medium" selection requirements described in Ref. [42]. The Z → ll candidates are required to be within the mass range 76 < m ll < 106 GeV and have p Z T > 15 GeV. This selection ensures that the contribution from multijet and other backgrounds is smaller than 1.5% (0.1%) for the dielectron (dimuon) channel, and is considered negligible. In total, these criteria select approximately 21 000 (28 000) Z → ee (Z → μμ) events in pp data, and 3400 (4100) events in Pb þ Pb data.
Each Z data event is assigned a series of weights, derived from simulation and data, to account for the trigger, reconstruction and selection efficiencies of its decay leptons. Individual lepton trigger efficiencies are determined directly in pp and Pb þ Pb data using tag-and-probe techniques [24,25], and are 0.70-0.80 for each muon and 0.75-0.95 for each electron. Reconstruction and selection efficiencies are determined using simulation and are 0.65-0.80 for muons and 0.65-0.95 for electrons. Although the efficiencies may vary substantially with the individual lepton p T , η, and ϕ, the resulting dependence on p Z T is weak due to the large Z mass and weak correlation between bosons and their decay leptons.
Charged-particle tracks are reconstructed from hits in the inner detector using an algorithm [43] which, in Pb þ Pb collisions, is optimized for the high-occupancy conditions [44]. They are required to meet several criteria intended to select primary charged particles [6]. All reconstructed tracks with p T > 1 GeV, jηj < 2.5 and Δϕ > 3π=4 are considered. The charged-particle yield is corrected for reconstruction and selection inefficiency on a per-track basis using a simulation-derived efficiency which varies from 0.6 to 0.8 depending on both detector occupancy and track kinematics. A small correction, typically 1%-2%, accounts for the contribution of reconstructed tracks not associated with primary particles. The p ch T resolution is found to have a negligible effect (≲0.3%) on the results and is not corrected for.
The contribution to the yield from UE particles in Pb þ Pb collisions is estimated using MB events and is statistically subtracted from the measured yields. For each Z event in data, 40-160 unique MB events are used for this estimation. These MB events are centrality matched to within 1% in peripheral events, decreasing to within 0.1% in central events. Furthermore, to match the azimuthal modulation of the UE, the elliptic flow angles [28] in the Z data event and in the matching MB event must match within π=16. The signal-to-background ratio varies strongly with p ch T , p Z T , and Pb þ Pb centrality, with a minimum of 5 × 10 −3 at the lowest p ch T and p Z T values in the most central events. In pp events, the UE is known to have larger activity in a Z event than in an ordinary MB pp collision [45,46], necessitating a different procedure. Here, the UE is determined in events with 1 < p Z T < 12 GeV in the azimuthal region perpendicular to the Z boson to avoid the contribution from jet particles.
The data are further corrected for bin migration resulting from the finite resolution in the p Z T measurement. This is evaluated by comparing the per-Z charged-particle yields, where the Z selection is made at the generator level, with those after reconstruction, and is typically a 2%-3% correction.
The primary sources of systematic uncertainty in the yield measurement are those affecting the Z boson reconstruction, those affecting the charged-particle selection, and those affecting the UE background estimation and subtraction. The uncertainties associated with the electron and muon energy scales are evaluated using a common set of uncertainties [42,47], and are typically negligible (≲1%) except at high p ch T . Those associated with lepton trigger and selection efficiency determination are smaller than the ones related to the energy scale. Several sources of trackingrelated uncertainty are considered, which are described in previous measurements of charged-particle fragmentation functions, and of which the largest is the sensitivity to the track selection criteria, which is 2%-3% [6,48].
The uncertainty in the determination of the UE background yield is evaluated by propagating the statistical uncertainty of the UE estimation in MB events. The sensitivity of the UE estimation to the matching criteria for the elliptic flow [27] angles between signal and MB events, or the additional requirement to match the triangular flow angles, are investigated. However, since these variations give statistically compatible results, they are not included. As a check of the background subtraction procedure, the full analysis is performed on simulated Z events overlaid with HIJING [49] Pb þ Pb background, and compared with the generator-level distributions. An absolute uncertainty in the background estimation of 0.3% is derived using this study.
Finally, an internal consistency check is performed by comparing the per-Z yields between the electron and muon decay channels. A difference was observed in the 15 < p Z T < 30 GeV selections and was included as an uncertainty of at most 4% in pp and 14% in central Pb þ Pb events.
For the yields at low p ch T and in central events, the uncertainty from the UE determination is dominant and can be as large as 30%. For yields at high p ch T and in lowermultiplicity events, the uncertainties associated with the track selection and the lepton energy scale are typically dominant, and as large as 5%. Uncertainty sources common to Pb þ Pb and pp are canceled in the I AA ratio when possible, such that the resulting measurement is dominated by uncertainties specific to Pb þ Pb events. In all cases, the statistical uncertainty in the I AA is larger than the total systematic uncertainty. Figure 1 presents the charged-particle yield per Z boson, in Pb þ Pb and pp events, as a function of p ch T , for the selection Δϕ > 3π=4. The yields in Pb þ Pb collisions are observed to be modified relative to those in pp collisions. T , for the selection Δϕ > 3π=4, reported for 15 < p Z T < 30 GeV, 30 < p Z T < 60 GeV, and p Z T > 60 GeV. Results are shown for pp events and the three centralities of Pb þ Pb events. These are offset horizontally around the bin centers, which are located between the 0%-10% and 10%-30% points, for visibility. The vertical bars and boxes correspond to the statistical and systematic uncertainties of the data. PHYSICAL REVIEW LETTERS 126, 072301 (2021) 072301-3 To better reveal the modification, Fig. 2 presents I AA values, the ratios of yields in Pb þ Pb events to those in pp events. The I AA values are suppressed below unity at large p ch T , with a systematically larger suppression in more central events and for lower p Z T selections. For p Z T > 60 GeV, the I AA values at low p ch T , less than 2-3 GeV, are significantly different than those at high p ch T , and typically greater than unity. Lower p Z T selections are compatible with a similar increase at low p ch T , although the uncertainties limit the significance of this enhancement. The suppression over a wide range of p ch T values, and the general enhancement of the I AA above unity at lower p ch T , are qualitatively similar to those observed in the ratios of jet fragmentation functions in photon-tagged events [12]. Figure 3 compares the I AA in 0%-10% Pb þ Pb events with the following theoretical calculations, where available, which use the same kinematic selections as the data: (1) a perturbative calculation within the framework of softcollinear effective field theory with Glauber gluons (SCET G ) in the soft-gluon-emission (energy-loss) limit, with jet-medium coupling g ¼ 2.0 AE 0.2 [50,51]; (2) the Hybrid Strong/Weak Coupling model [52], which combines initial production using PYTHIA 8 with a parameterization of energy loss derived from holographic methods, including backreaction effects; (3) JEWEL, an MC event generator which simulates QCD jet evolution in heavy-ion collisions, including radiative and elastic energy loss processes, and configured to include medium recoils [53]; and (4) a coupled linearized Boltzmann transport (COLBT) and hydrodynamics model [54,55], which includes jet-induced medium excitations. All models qualitatively reproduce the degree of suppression at large p ch T , greater than 10 GeV. The Hybrid model, JEWEL and COLBT qualitatively capture the increase at low p ch T . For these three models, removing the backreaction, medium recoils, and jet-induced medium excitations, respectively, results in a significant underprediction of the data in this region. Several of these models also capture the relative difference in the I AA between the three p Z T selections. A full evaluation of theoretical uncertainties is needed to further discriminate between the mechanisms of energy loss and medium response in the data.   FIG. 2. Ratio of the charged-particle yield in Pb þ Pb collisions to that in pp collisions, I AA , as a function of charged-particle p ch T , for the selection Δϕ > 3π=4. The vertical bars and boxes correspond to the statistical and systematic uncertainties of the data. The 0%-10% and 30%-80% data are offset horizontally for visibility. In conclusion, this Letter presents a measurement of charged-particle yields produced in the azimuthal direction opposite to a Z boson with p T > 15 GeV. The measurement is performed using 260 pb −1 of pp and up to 1.7 nb −1 of Pb þ Pb collision data at 5.02 TeV with the ATLAS detector at the Large Hadron Collider. The per-Z yields are systematically modified in Pb þ Pb collisions compared with pp collisions due to the interactions between the parton shower and the hot and dense QGP medium. The charged-particle p T distribution in Pb þ Pb collisions is softer than that in pp collisions, with a suppression at high p ch T and an enhancement at low p ch T . The degree of modification varies with Pb þ Pb event centrality, consistent with a larger and hotter QGP being created in more central events. At high p Z T , the modification pattern is qualitatively similar to that observed in measurements of photon-tagged jet fragmentation functions. In addition to the particular theoretical comparisons presented here, the data will allow systematic tests of models across centrality and p Z T selections. The data can also test energy loss models for low-p T partons that are otherwise difficult to access experimentally at the LHC, but which are valuable for direct comparison to future measurements at RHIC.  [42] ATLAS Collaboration, Muon reconstruction performance of the ATLAS detector in proton-proton collision data at ffiffi ffi s p ¼ 13 TeV, Eur. Phys. J. C 76, 292 (2016).