ϒ production in p–Pb collisions at √sNN = 8.16 TeV

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY 4.0 https://creativecommons.org/licenses/by/4.0/ ϒ production in p–Pb collisions at √sNN = 8.16 TeV © 2020 European Organization for Nuclear Research. Published by Elsevier B.V. Published version ALICE Collaboration ALICE Collaboration. (2020). ϒ production in p–Pb collisions at √sNN = 8.16 TeV. Physics Letters B, 806, Article 135486. https://doi.org/10.1016/j.physletb.2020.135486 2020 Physics Letters B 806 (2020) 135486 Contents lists available at ScienceDirect Physics Letters B www.elsevier.com/locate/physletb ϒproduction in p–Pb collisions at √sNN =8.16 TeV .ALICE Collaboration a r t i c l e i n f o a b s t r a c t Article history: Received 16 November 2019 Received in revised form 29 April 2020 Accepted 11 May 2020 Available online 20 May 2020 Editor: D.F. Geesaman ϒproduction in p–Pb interactions is studied at the centre-of-mass energy per nucleon–nucleon collision √sNN =8.16 TeV with the ALICE detector at the CERN LHC. The measurement is performed reconstructing bottomonium resonances via their dimuon decay channel, in the centre-of-mass rapidity intervals 2.03 <ycms <3.53 and −4.46 <ycms <−2.96, down to zero transverse momentum. In this work, results on the ϒ(1S) production cross section as a function of rapidity and transverse momentum are presented. The corresponding nuclear modification factor shows a suppression of the ϒ(1S) yields with respect to pp collisions, both at forward and backward rapidity. This suppression is stronger in the low transverse momentum region and shows no significant dependence on the centrality of the interactions. Furthermore, the ϒ(2S) nuclear modification factor is evaluated, suggesting a suppression similar to that of the ϒ(1S). A first measurement of the ϒ(3S) has also been performed. Finally, results are compared with previous ALICE measurements in p–Pb collisions at √sNN =5.02 TeV and with theoretical calculations. ©2020 European Organization for Nuclear Research. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Funded by SCOAP3. 1. Introduction Quarkonium resonances, i.e. bound states of a heavy quark (Q) and anti-quark (Q), are well-known probes of the formation of a quark–gluon plasma (QGP) which can occur in heavy-ions collisions. The high colour-charge density reached in such a medium can, in fact, screen the binding force between the Q and Q, leading to a temperature-dependent melting of the quarkonium states according to their binding energies [1]. A suppression of bottomonium resonances, the bound states formed by b and bquarks, was observed in Pb–Pb collisions, at the LHC energies of √sNN =2.76 TeV and √sNN =5.02 TeV by the ALICE [2,3] and CMS [4–6] experiments. All the ϒresonances show a reduction in their production yields compared to pp interactions at the same centre-of-mass energy, scaled by the number of nucleon– nucleon collisions. Furthermore, the magnitude of the suppression is significantly different for the three resonances and it increases from the tightly bound ϒ(1S) to the loosely bound ϒ(3S) [4–6], as expected in a sequential suppression scenario, with the binding energies of the ϒstates ranging between ∼1 GeV for the ϒ(1S) to ∼0.2 GeV for the ϒ(3S) [7]. Modifications to the bottomonium production might also be induced by cold nuclear matter (CNM) mechanisms not related to the formation of the QGP. The modification of the quark and gluon structure functions for nucleons inside nuclei, modelled either via nuclear parton distribution functions E-mail address: alice -publications @cern .ch. (nPDFs) [8–11]or through a Color Glass Condensate effective theory [12], or the coherent energy loss of the QQpair during its path through the cold nuclear medium [13]are examples of CNM effects which can influence quarkonium production [14]. The size of these effects is usually assessed in proton–nucleus collisions. These interactions also allow for the investigation of additional final state mechanisms, which can modify the production in particular of the more loosely bound resonances [15–17]. ALICE has published results on the modification of the ϒ(1S) production yields as a function of the centre-of-mass rapidity (ycms) using the 2013 p–Pb collisions data sample at √sNN = 5.02 TeV [18]. The size of the observed suppression was found to be similar in the forward and backward rapidity regions. Theoretical calculations based on the aforementioned CNM mechanisms fairly describe the forward-ycms measurements, while they slightly overestimate the results obtained at backward rapidity. Furthermore, the measurement of the ϒ(2S) to ϒ(1S) ratio [18], ϒ(2S)/ϒ(1S), was consistent, albeit within large uncertainties, with the one obtained in pp collisions [19], suggesting CNM effects of the same size on the two resonances both at forward and backward rapidity. Consistent results were also obtained by the LHCb experiment [20]in a similar kinematic region. However, it should be noted that ATLAS [21] and CMS [22] measurements of ϒ(2S)/ϒ(1S) at midrapidity suggest a stronger suppression of the ϒ(2S) with respect to the ϒ(1S) state, as expected if final state effects are at play [15]. In 2016, the LHC delivered p–Pb collisions at √sNN =8.16 TeV. The increase both in integrated luminosity, about a factor of 2 https://doi.org/10.1016/j.physletb.2020.135486 0370-2693/©2020 European Organization for Nuclear Research. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Funded by SCOAP3. 2ALICE Collaboration / Physics Letters B 806 (2020) 135486 larger than the one collected in 2013, and in the bottomonium production cross section, due to the higher centre-of-mass energy, allows a more detailed study of the production of the ϒstates. In this paper, results on the ϒ(1S) production as a function of ycms, transverse momentum (pT) and centrality of the collisions will be discussed and compared with the measurements performed in p–Pb collisions at √sNN =5.02 TeV and with theoretical calculations. A comparison of the ϒ(2S) and ϒ(3S) to ϒ(1S) production yields and nuclear modification factors, integrated over ycms, pTand centrality, will also be presented. Finally, the results will be compared with the corresponding measurements obtained by LHCb at the same energy [23]. It should be noted that all the presented results refer to the ϒinclusive production, i.e. to ϒeither produced directly or coming from the feed-down of higher-mass excited states. 2. Experimental apparatus and data sample A detailed description of the ALICE apparatus and performance can be found in [24,25]. The forward muon spectrometer [26]is the main detector used in this analysis. It consists of five tracking stations made of two planes of Cathode Pad Chambers each, followed by two trigger stations each one composed by two planes of Resistive Plate Chambers. A 10 interaction-length (λI) absorber, placed in front of the tracking system, filters out most of the hadrons produced in the collisions. Low-momentum muons and hadrons escaping the first absorber are stopped by a second 7.2 λI-thick iron wall, placed in front of the trigger stations. The momentum of the particles is evaluated by measuring their curvature in a dipole magnet with a 3 T×m integrated field. The muon spectrometer measures muons in the pseudorapidity interval −4 < η<−2.5in the laboratory reference frame. It also provides single and unlikeor like-sign dimuon triggers based on the detection in the trigger system of one or two muons, respectively, having a transverse momentum higher than a programmable threshold set to pT,μ=0.5GeV/c. This threshold is not sharp and the single muon trigger efficiency reaches a plateau value of ∼98% at about pT,μ∼1.5GeV/c. The primary interaction vertex of the collision is reconstructed using the two innermost layers of the Inner Tracking System (Silicon Pixel Detector, SPD) [27], extending over the pseudorapidity intervals |η| <2 and |η| <1.4, respectively. The V0 detector [28], composed of two sets of scintillators covering the pseudorapidity intervals 2.8 <η<5.1 and −3.7 <η<−1.7, provides the luminosity measurement, which can also be obtained independently from the information of the T0 Cherenkov detectors [29], covering the regions 4.6 <η<4.9 and −3.3 <η<−3. The V0 detector is also used to provide the minimum bias (MB) trigger, defined by the coincidence of signals in the two sets of scintillators. The trigger condition used in this analysis is based on the coincidence of the MB trigger with the unlike-sign dimuon one (μμ-MB). The removal of beam-induced background is based on the timing information provided by the V0 and by two sets of Zero Degree Calorimeters (ZDC) [30]placed at ±112.5 m from the interaction point, along the beamline. The ZDCs are also used for the centrality estimation as it will be discussed in Sec. 3. Finally, for the study of the ϒproduction as a function of the centrality of the collisions, pile-up events in which two or more interactions occur in the same colliding bunch are removed using the information from SPD and V0. Further selection criteria, commonly adopted in the ALICE quarkonium analyses (see e.g. [18,31]), are applied to the muon tracks forming the dimuon pair. Muon tracks must have a pseudorapidity value in the range −4 <ημ<−2.5, corresponding to the muon spectrometer acceptance, and they should point to the interaction vertex to remove fake tracks and particles not directly produced in beam–beam interactions. Their transverse coordinate at the end of the front absorber (Rabs) must be within 17.6cm<Rabs <89.5cm, to remove muons not passing the homogeneous region of the absorber. Finally, tracks reconstructed in the tracking chambers of the muon spectrometer should match the track segments reconstructed in the trigger system. This matching request helps to further reject hadron contamination and ensures that the reconstructed muons fulfill the trigger condition. The data were collected with two beam configurations obtained by inverting the directions of the proton and Pb beams circulating inside the LHC. In this way it was possible to cover both a forward (2.03 <ycms <3.53) and a backward (−4.46 <ycms <−2.96) dimuon rapidity interval, where the positive (negative) ycms refers to the proton (Pb) beam going towards the muon spectrometer. The collected integrated luminosities for the corresponding data samples, referred to as p–Pb (forward rapidity) and Pb–p (backward rapidity) in the following, are LpPb int =8.4 ±0.2nb −1and LPbp int =12.8 ±0.3nb −1[32]. 3. Data analysis The results presented in this paper are based on an analysis procedure similar to the one described in [18]for the study of the ϒproduction in p–Pb collisions at √sNN =5.02 TeV. The ϒ(1S), ϒ(2S) and ϒ(3S) production cross sections, corrected by the branching ratio for the decay in a muon pair (B.R.ϒ→μ+μ−), are obtained, for a given (ycms, pT) interval, as d2σϒ pPb dycmsdpT=Nϒ LpPb int ×(A×ε)×ycms ×pT×B.R.ϒ→μ+μ− , (1) where Nϒis the number of signal counts and (A ×ε) is the corresponding acceptance and efficiency correction in the kinematic bin under study, while the branching ratios are (2.48 ±0.05)% for ϒ(1S), (1.93 ±0.17)% for ϒ(2S) and (2.18 ±0.21)% for ϒ(3S) [33]. The number of ϒ(nS)is obtained by fitting the unlike-sign dimuon invariant mass spectrum with a combination of signal shapes to describe the ϒresonances and an empirical function to model the background. More in detail, the background is described by several combinations of exponential and polynomial functions or by a Gaussian function with a mass-dependent width. For the resonance shapes, extended Crystal Ball functions [34], with power-law tails on the right and left sides of the mass peak are used. Alternatively, pseudo-Gaussian functions with a massdependent width are also adopted [34]. The same signal shape is chosen for all the ϒstates. The mass of the ϒ(1S) and its width σϒ(1S)are free parameters of the fit, while the mass and the width of the ϒ(2S) and ϒ(3S) states are bound to those of the ϒ(1S) in the following way: mϒ(nS)=mϒ(1S)+(mPDG ϒ(nS)−mPDG ϒ(1S)) and σϒ(nS)=σϒ(1S)×σMC ϒ(nS)/σMC ϒ(1S). The mass value mPDG ϒ(nS)is taken from [33] and σMC ϒ(nS)is the width of the resonance as evaluated from a fit, with the aforementioned signal functions, to the spectrum obtained from the Monte Carlo (MC) simulation also used for the (A ×ε) correction. Due to the signal-over-background ratio of the order of ∼0.7 (∼1) in p–Pb (Pb–p), measured in a 3σ region around the ϒ(1S) mass, the non-Gaussian tails of the extended Crystal Ball function can not be kept as free parameters of the fits. Hence, they are tuned on pp data at √s=13 TeV, the largest data sample collected by ALICE so far, or, alternatively, on p–Pb or pp MC simulations at √sNN =8.16 TeV and √s=8TeV, respectively. The same tails are adopted for the ϒ(2S) and ϒ(3S) mass shapes. Examples of the fit to the invariant mass spectrum, for both the p–Pb and Pb–p samples, are shown in Fig. 1. ALICE Collaboration / Physics Letters B 806 (2020) 135486 3 Fig. 1. Invariant mass spectra of unlike-sign dimuons, integrated over pT, for Pb–p (left panel) and p–Pb (right panel) collisions. The shapes of the ϒ(1S), ϒ(2S) and ϒ(3S) resonances are shown (dash-dotted lines), together with the background function (dashed line) and the total fit (solid line). The number of ϒcandidates, Nϒ, is evaluated as the average of the values obtained by varying the signal and background functions as well as the fitting intervals (6 GeV/c2<mμμ <13 GeV/c2 or 7GeV/ c2<mμμ <12 GeV/c2). The statistical uncertainties are calculated as the average of the statistical uncertainties over the various fits and the standard deviation of the distribution of the Nϒvalues provides the systematic uncertainties on the signal extraction. For the ϒ(2S) and ϒ(3S) cases, an additional contribution to the systematic uncertainty is included, to account for possible variations of their width with respect to that of the ϒ(1S). In particular, their widths are allowed to vary between a minimum value σϒ(1S)and a maximum value σϒ(1S)×σMC ϒ(nS)/σMC ϒ(1S), where the ratio σMC ϒ(nS)/σMC ϒ(1S)is obtained from MC simulations alternative to the ones used for the (A ×ε) correction, i.e. based on different ϒ kinematic input shapes, as it will be discussed later on. A further 5% systematic uncertainty is also included to account for possible residual discrepancies between the detector resolution in MC and in the data. The total number of ϒ(1S), integrated over the full kinematic range, amounts to Nϒ(1S)=909 ±62 (stat.) ±58(syst.) and Nϒ(1S)=918 ±55 (stat.) ±51(syst.) for the forward and backward-rapidity regions, respectively. Corresponding values for ϒ(2S) are Nϒ(2S)=192 ±39 (stat.) ±17(syst.) and Nϒ(2S)=194 ± 34 (stat.) ±16(syst.), while for the ϒ(3S) the values are Nϒ(3S)= 48 ±36 (stat.) ±8(syst.) and Nϒ(3S)=95 ±30 (stat.) ±12(syst.). The systematic uncertainty, amounting to ∼6% for the ϒ(1S) and ∼8% for the ϒ(2S), is dominated by the choice of the tail parameters in the fit functions and, in the ϒ(2S) case, also by the allowed range of variation for the σϒ(2S). In the ϒ(3S) case, the systematic uncertainties are slightly larger, amounting to ∼17% at forward rapidity and ∼12% at backward rapidity. For pT-or ycms-differential ϒ(1S) studies, the systematic uncertainties have a similar size, reaching ∼15% only in the highest pTbin (8 GeV/c< pT<15 GeV/c). The acceptance and efficiency correction is calculated in a MC simulation, based on the GEANT3 transport code [35]. The MC simulation is performed on a run-by-run basis to closely follow the evolution of the performance of the detectors during the data taking. The ϒ(1S) are generated using rapidity and transverse momentum distributions tuned on p–Pb or Pb–p data at √sNN = 8.16 TeV, through an iterative procedure [31]. The pTand ycms integrated (A ×ε) amounts to 0.300 ±0.006 for the ϒ(1S) at forward rapidity and 0.273 ±0.007 at backward rapidity, where the quoted uncertainties are systematic, the statistical uncertainties being negligible. The lower (A ×ε) values measured in the Pb–p period, with respect to the p–Pb one, are due to detector instabilities which affected temporarily the behaviour of two tracking chambers. The limited size of the data sample do not allow for a similar tuning of the pTand ycms distributions on data for the ϒ(2S) and ϒ(3S) resonances, hence the same shapes as for the ϒ(1S) are used. The resulting (A ×ε) values show a negligible difference with respect to the ϒ(1S) ones. The systematic uncertainties on (A ×ε) include contributions related to the choice of the MC pTand ycms input distributions for the ϒstates and to the evaluation of the tracking and trigger efficiencies. The systematic uncertainties associated to the MC ϒinput shapes are evaluated as the maximum difference between the (A ×ε) evaluated with the aforementioned MC tuned on data and the values extracted from alternative MC samples based on pTand ycms ϒdistributions either measured by the LHCb experiment in pp collisions at √s=8TeV[36]or obtained from existing CDF and LHC pp measurements [37–39]via a procedure similar to the one described in [40]. Nuclear shadowing is also included to account for its influence on the bottomonium kinematic distributions. These systematic uncertainties for the three resonances vary between 1% and 1.8%. They have a negligible pT-dependence, while they reach up to 4% at the edges of the rapidity intervals. The systematic uncertainty on the trigger efficiency consists of two contributions, one related to the evaluation of the intrinsic efficiency of each muon-trigger chamber (1%) and one to small differences between the trigger response function estimated via data and MC (0.6% in p–Pb and 0.2% in Pb–p, when integrating over ycms and pT). This last source of uncertainty is below 1% also for the pTor ycms-differential studies. The systematic uncertainty associated to the tracking efficiency is evaluated comparing the dimuon tracking efficiencies computed both in data and MC. These efficiencies are computed combining the efficiency of each single muon-tracking chamber, obtained relying on the redundancy of the tracking system. The resulting systematic uncertainties amount to 1% for p–Pb and 2% for Pb–p, for both the ycms and pTdifferential studies and for results integrated over the kinematic domain. Finally, an additional 1% systematic uncertainty on the choice of the χ2cut on the matching between the tracks reconstructed in the tracking and in the trigger systems is included. The systematic uncertainties associated to the trigger, tracking and matching efficiencies are considered to be identical for both the ϒ(1S) and ϒ(2S) resonances. The integrated luminosities are obtained as Lint =NMB/σMB. The number of equivalent minimum bias events, NMB, is evaluated by multiplying the number of events collected with the μμ-MB 4ALICE Collaboration / Physics Letters B 806 (2020) 135486 trigger by a factor Fnorm, corresponding to the inverse of the probability of having a triggered dimuon in a MB event [31]. This quantity is computed, run by run, as the ratio between the number of collected MB triggers and the number of times the dimuon trigger condition is verified in the MB trigger sample. Once averaged over all the runs, considering as weight the number of μμ-MB triggers in each run, Fnorm amounts to 679 ±7at forward rapidity and 372 ±4at backward rapidity. The quoted uncertainty (1%) is systematic and accounts for differences coming from an alternative evaluation method, based on the information provided by the level-0 trigger scalers, as detailed in [41]. The V0-based MB cross section (σMB) is measured from a van der Meer scan, and it amounts to 2.09 ±0.04 b for the p–Pb configuration and 2.10 ±0.04 b for the Pb–p one [32]. In the luminosity systematic uncertainty quoted in Table 1, the uncertainties on Fnorm and σMB are combined, together with a 1.1% (0.6%) contribution due to the difference between the luminosities obtained with the V0 and T0 detectors in the p–Pb (Pb–p) configurations [32]. The nuclear effects on the ϒproduction are studied comparing the corresponding p–Pb production cross section to the one measured in pp collisions, d2σϒ pp/dycmsdpT, obtained at the same centre-of-mass energy and scaled by the atomic mass number of the Pb nucleus (APb =208), through the so-called nuclear modification factor RpPb, defined as RpPb = d2σϒ pPb/dycmsdpT APb ×d2σϒ pp/dycmsdpT.(2) The proton–proton reference is based on the LHCb measurements of the bottomonium production cross section in pp collisions at √s=8TeV[36], in −4.5 <ycms <−2.5and 2 < ycms <4, corrected by a factor to account for the slightly different centre-of-mass energies of the interactions. This correction factor is evaluated interpolating the LHCb measurements at √s=7, 8 and 13 TeV [36,42], as detailed in [43]. It amounts to 1.02 for both the ϒ(1S) and ϒ(2S), showing a negligible ycms dependence and varying by 1% from low to high pT. A systematic uncertainty on the determination of this factor (1%) is assigned, based on the choice of the different functions used for the energy-interpolation. The ϒproduction cross sections in pp collisions at √s= 8 TeV are also measured by ALICE [44]. The results show good agreement with the corresponding LHCb values, but unlike the LHCb measurements, they cover a slightly narrower rapidity region, 2.5 <ycms <4, which does not match the rapidity coverage of the p–Pb measurements. The σϒ(1S) pp cross sections, integrated over pTand ycms, are 98.5 ±0.1 (stat.) ±3.4 (syst.) nb in the range 2.03 <ycms <3.53 and 62.0 ±0.1 (stat.) ±2.1 (syst.) nb in the range −4.46 <ycms < −2.96. The corresponding cross sections for the ϒ(2S) are about a factor 3 smaller, being σϒ(2S) pp =31.9 ±0.1 (stat.) ±2.9(syst.) nb at forward rapidity and 19.7 ±0.05 (stat.) ±1.8(syst.) nb at backward rapidity. The ϒ(3S) production cross sections are σϒ(3S) pp = 12.9 ±0.1 (stat.) ±1.3(syst.) nb at forward rapidity and 8.3 ± 0.1 (stat.) ±0.8(syst.) nb at backward rapidity. The large data sample collected in p–Pb collisions in 2016 allows the ϒ(1S) production also to be studied as a function of the collision centrality. The centrality determination is based on a hybrid model, as discussed in detail in [45]. In this approach, the centrality is determined by measuring the energy released in the ZDC positioned in the Pb-going direction. For each ZDC-selected centrality class, the average number of collisions Ncollis obtained as Ncoll =Npart-1, assuming the charged particle multiplicity measured at midrapidity is proportional to the number of participant nucleons, Npart. The centrality classes used in this analysis correspond to 2–20%, 20–40%, 40–60% and 60–90% of the MB cross section. The 0–2% most central collisions are excluded from this analysis because the fraction of events coming from pile-up in the ZDC is large in this centrality interval and a residual contamination might still be present in spite of the applied pile-up rejection cuts [46]. For centrality studies, the modification induced by the nuclear matter on the ϒ(1S) production is quantified through the nuclear modification factor denoted by QpPb, to be distinguished from RpPb since potential biases from the centrality estimation, unrelated to nuclear effects, might be present [45]. The QpPb is defined as QpPb =Nϒ B.R.ϒ→μ+μ−×NMB ×(A×ε)×TpPb×σϒ pp .(3) The quantities entering Eq. (3)are evaluated according to the previously discussed procedure, with few minor differences. When extracting the ϒ(1S) signal, for example, no significant variation of the ϒ(1S) width as a function of the collision centrality is foreseen. Hence for centrality studies, the ϒ(1S) width is fixed to the value obtained in the fit to the centrality-integrated invariant mass spectrum. The uncertainty associated to the choice of the width is accounted for in the evaluation of the systematic uncertainty on the signal extraction. No significant centrality dependence is expected for the (A ×ε) either, so the centrality-integrated values are also used for all the centrality classes. To evaluate the number of MB events in each centrality class i, Fi norm is obtained from the centrality-integrated quantity scaled by the ratio of the number of minimum bias and dimuon-triggered events in each centrality interval with respect to the corresponding centrality integrated quantities, (Ni MB/NMB)/(Ni μμ−MB/Nμμ−MB). Alternatively, Fi norm is computed directly for each centrality class and a further 1% difference between the two approaches is included in the systematic uncertainty. The statistical uncertainty on Fi norm is negligible. Finally, TpPbis the centrality-dependent average nuclear thickness function, computed with the Glauber framework [45,47]. The systematic uncertainties entering the cross section and nuclear modification factor evaluation are summarised in Table 1. When RpPb is computed as a function of pTor ycms, the systematic uncertainties on the signal extraction, tracking, trigger and matching efficiencies, MC input shapes and a fraction of the uncertainty on the pp reference are considered as bin-by-bin uncorrelated. On the contrary, the correlated contributions to the pp reference and the luminosity uncertainties, which are common to the p–Pb or Pb–p systems, are considered as correlated over pTor ycms. In the QpPb evaluation, the uncertainties on signal extraction, on the MC input shapes and on TpPbdepend on the centrality of the collision, while the other uncertainties are common to all classes and, therefore, considered as correlated over centrality. Even if most central events are not included in this analysis, a further 2% centrality-uncorrelated systematic uncertainty is assigned to the QpPb values, to account for residual pile-up which might still introduce a bias in the measurement. This systematic uncertainty is evaluated by comparing the expected pile-up fraction, computed from the pile-up probability associated to the observed interaction rate, and the amount of pile-up events removed by the event selection procedure. For the ϒ(2S) and ϒ(3S) studies, similar values of the systematic uncertainties are obtained, the main difference being the larger signal extraction uncertainties. 4. Results The inclusive ϒ(1S) production cross sections are evaluated in the rapidity regions 2.03 <ycms <3.53 and −4.46 <ycms <−2.96 and their values, computed according to Eq. (1), are: ALICE Collaboration / Physics Letters B 806 (2020) 135486 5 Table 1 Systematic uncertainties, in percentage, on the three ϒcross sections and nuclear modification factors for both p–Pb and Pb–p collisions. Ranges in parentheses refer to the maximum variation as a function of centrality, ycms or pT. When no ranges are specified, the quoted values are valid for both the integrated and the differential measurements. Error type I means that the uncertainties are correlated over pTor ycms, while error type II refers to uncertainties correlated versus centrality. If no error type is specified, the uncertainties are considered as uncorrelated. The uncertainties on the pp reference and luminosity result from the combination of ycms-uncorrelated and correlated contributions. For the systematic uncertainty on the luminosity determination, the two terms, defined according to [32], are separately quoted in the table, but combined when results are shown in the figures. Uncertainties on the B.R. are taken from [33]. Sources ϒ(1S) ϒ(2S) ϒ(3S) p–Pb Pb–p p–Pb Pb–p p–Pb Pb–p Signal extraction 6.4 (5.1–15.9) 5.7 (5.5–8.5) 8.8 8.4 17.4 12.6 Trigger efficiency (II) 1.2 (1.1–1.3) 1.0 (1.0–1.1) 1.2 1.0 1.2 1.0 Tracking efficiency (II) 1.0 2.0 1.0 2.0 1.0 2.0 Matching efficiency (II) 1.0 1.0 1.0 1.0 1.0 1.0 MC inputs 1.0 (0.5–4.0) 1.0 (0.4–4.0) 1.3 1.6 1.4 1.8 pp reference (II) 0.2 (0.1–0.4) 0.2 (0.1–0.4) 0.2 0.3 0.2 0.2 pp reference (I,II) 2.8 2.8 2.8 LpPb int (II) 2.1 2.2 2.1 2.2 2.1 2.2 LpPb int (I,II) 0.5 0.7 0.5 0.7 0.5 0.7 Pile-up 2.0 2.0 – – TpPb2.1–5.8 – – B.R. (I) 2.0 8.8 9.6 σϒ(1S) pPb (2.03 <ycms <3.53) =14.5±1.0(stat.) ±1.0(uncor.syst.) ±0.3(cor.syst.) μb, σϒ(1S) pPb (−4.46 <ycms <−2.96) =10.5±0.6(stat.) ±0.7(uncor.syst.) ±0.2(cor.syst.) μb. The corresponding values for the ϒ(2S) production cross sections are: σϒ(2S) pPb (2.03 <ycms <3.53) =3.9±0.8(stat.) ±0.4(uncor.syst.) ±0.3(cor.syst.) μb, σϒ(2S) pPb (−4.46 <ycms <−2.96) =2.8±0.5(stat.) ±0.3(uncor.syst.) ±0.3(cor.syst.) μb, and for the ϒ(3S) are: σϒ(3S) pPb (2.03 <ycms <3.53) =0.87 ±0.66 (stat.) ±0.15 (uncor.syst.) ±0.08 (cor.syst.) μb, σϒ(3S) pPb (−4.46 <ycms <−2.96) =1.24 ±0.39 (stat.) ±0.15 (uncor.syst.) ±0.12 (cor.syst.) μb. The systematic uncertainties have two terms, one correlated and one uncorrelated as a function of rapidity. The data collected in p–Pb collisions at √sNN =8.16 TeV allow for the measurement of the ϒ(1S) production cross sections differentially in ycms bins or in pTintervals, up to pT<15 GeV/c. The resulting cross sections are shown in Fig. 2as a function of rapidity, integrated over transverse momentum, and in Fig. 3, as a function of pT, in the forwardand backward-rapidity regions. In these figures, as in all the following ones, the statistical uncertainties are shown as vertical error bars, while the systematic uncertainties are represented as boxes around the points. The horizontal error bars correspond to the ycms or pTbin widths. The cross sections evaluated at forward and backward rapidities are compared with the pp ones, obtained through the aforementioned interpolation procedure, scaled by the Pb atomic mass number. The comparison shows that in the forward-rapidity region the ϒ(1S) cross sections are smaller than the pp ones, in particular at low pT, suggesting the presence of CNM effects at play in p–Pb collisions. On the contrary, in the backward-rapidity range the pp and the p–Pb cross Fig. 2. ϒ(1S), ϒ(2S) and ϒ(3S) differential cross sections as a function of ycms in p–Pb collisions at √sNN =8.16 TeV. The corresponding pp reference cross sections, obtained through the procedure described in Sec. 3and scaled by APb, are shown as bands. Fig. 3. ϒ(1S) differential cross section as a function of pT, at forward (closed symbols) and backward (open symbols) rapidity, at √sNN =8.16 TeV. The pp reference cross section, obtained through the procedure described in Sec. 3and scaled by APb, is shown as a band. 6ALICE Collaboration / Physics Letters B 806 (2020) 135486 Fig. 4. Ratio of ϒ(nS)over ϒ(1S) yields in p–Pb collisions at √sNN =8.16 TeV and in pp collisions at √s= 8 TeV [36]. sections are closer and nuclear effects seem to have a less prominent role. The limited available data sample allows for the evaluation of the ϒ(2S) and ϒ(3S) cross sections in the forward and backwardrapidity regions only integrating over the corresponding ycms and pTranges, as shown in Fig. 2. A suppression with respect to the corresponding pp reference cross sections, scaled by APb, is observed. Given the relatively small mass difference between the ϒ(1S) and ϒ(2S) (or ϒ(3S)) resonances, most of the systematic uncertainties, except those on the signal extraction and on the choice of the pTand ycms-input shapes used in the MC, cancel in the ratio of the resonance yields, multiplied by their branching ratios, defined as [ϒ(nS)/ϒ(1S)]pPb =Nϒ(nS)/(A×ε)ϒ(nS) Nϒ(1S)/(A×ε)ϒ(1S) . The values of the ϒ(2S) over ϒ(1S) ratio, obtained at forward and backward rapidity, are similar: [ϒ(2S)/ϒ(1S)]pPb(2.03 <ycms <3.53) =0.21 ±0.05 (stat.) ±0.02 (syst.), [ϒ(2S)/ϒ(1S)]pPb(−4.46 <ycms <−2.96) =0.21 ±0.04 (stat.) ±0.01 (syst.). As shown in Fig. 4, the ratio [ϒ(2S)/ϒ(1S)]pPb at √sNN =8.16 TeV is compatible, within uncertainties, with the results obtained by the LHCb Collaboration in pp collisions at √s= 8 TeV [36], in a slightly wider kinematic range (2 <ycms <4.5, pT<15 GeV/c). Similar conclusions can be obtained from the comparison of the ϒ(3S) over ϒ(1S) ratio, also shown in Fig. 4. The corresponding values at forward and backward rapidity are: [ϒ(3S)/ϒ(1S)]pPb(2.03 <ycms <3.53) =0.053 ±0.039 (stat.) ±0.007 (syst.), [ϒ(3S)/ϒ(1S)]pPb(−4.46 <ycms <−2.96) =0.102 ±0.032 (stat.) ±0.009 (syst.). The size of nuclear effects in p–Pb collisions can be better quantified through the nuclear modification factor defined in Eq. (2). The numerical values for the ϒ(1S) RpPb in the forwardand in the backward-rapidity regions, integrating over pT, are: Fig. 5. ϒ(1S) RpPb values at √sNN =8.16 TeV compared to those obtained at √sNN =5.02 TeV in the same ycms interval [18]. All systematic uncertainties are considered as uncorrelated between the results at √sNN =8.16 TeV and √sNN = 5.02 TeV. The RpPb values at the two energies are slightly displaced horizontally to improve visibility. Rϒ(1S) pPb (2.03 <ycms <3.53) =0.71 ±0.05 (stat.) ±0.05 (uncor.syst.) ±0.02 (cor.syst.), Rϒ(1S) pPb (−4.46 <ycms <−2.96) =0.81 ±0.05 (stat.) ±0.05 (uncor.syst.) ±0.02 (cor.syst.), where (uncor. syst.) and (cor. syst.) refer to uncorrelated and correlated systematic uncertainties as a function of rapidity. The measured RpPb values, shown in Fig. 5, indicate a suppression of the ϒ(1S) production in p–Pb collisions, with respect to the one in pp collisions, both at forward and backward rapidity, with a slightly stronger suppression at forward ycms. The RpPb is found to be 4.0σand 2.4σbelow unity in p–Pb and Pb–p collisions, respectively. The results are compatible with the corresponding RpPb values measured in p–Pb collisions at √sNN =5.02 TeV [18], also shown in Fig. 5. From the comparison between the results obtained at the two energies, an improvement in the precision of the ϒ(1S) RpPb measurements at √sNN =8.16 TeV can be noticed, given the reduced size of the statistical and systematic uncertainties. The improvement of the latter contribution is mainly related to the reduction in the uncertainties associated to the tracking efficiencies and to refinements in the determination of the pp reference [18]. The rapidity dependence of the ϒ(1S) RpPb, explored in narrower ycms intervals, is shown in Fig. 6, confirming the suppression already observed in the ycms-integrated case. The results are also compared with the ϒ(1S) LHCb measurements [23]at the same centre-of-mass energy and in slightly wider kinematic ranges (−4.5 <ycms <−2.5and 2 <ycms <4, pT<25 GeV/c). Fair agreement between the two sets of results can be seen. The pTdependence of the ϒ(1S) RpPb is shown in Fig. 7. A slight decrease of the ϒ(1S) nuclear modification factor, with decreasing pT, is observed. The behaviour is similar both at backward and forward rapidities. The ycms and pTdependence of the ϒ(1S) RpPb are compared, in Fig. 6and Fig. 7, to several models (referred in the following as nuclear shadowing models), based on EPS09 [8], nCTEQ15 [10] or EPPS16 [9]sets of nuclear parton distribution functions. The EPS09 next-to-leading order (NLO) parametrisation is combined with a NLO Colour Evaporation Model (CEM) [48], which describes the ϒproduction. The corresponding uncertainty bands, shown in Fig. 6and Fig. 7, are dominated by the uncertainties of the EPS09 parametrisation. The nCTEQ15 and the EPPS16 NLO nPDFs ALICE Collaboration / Physics Letters B 806 (2020) 135486 7 Fig. 6. ϒ(1S) RpPb values at √sNN =8.16 TeV compared with the corresponding LHCb results [23], as a function of ycms. The RpPb values are also compared to model calculations based on several implementations of nuclear shadowing (EPS09 NLO [8,14,48], EPPS16 and nCTEQ15 [9–11,49–51]) and on parton coherent energy loss predictions, with or without the inclusion of the EPS09 shadowing contribution [13,14]. A theoretical model including a shadowing contribution based on nCTEQ15 nPDFs on top of a suppression induced by comover interactions [15,52] is also shown. For the LHCb results, the vertical error bars represent the quadratic sum of the statistical and systematic uncertainties. sets are implemented following the Bayesian reweighting procedure described in [11,49–51]. The uncertainty bands, in this case, represent the convolution of the uncertainties on the nPDFs sets and those on the factorisation scales. It can be observed that the shadowing calculations describe fairly well the pTand ycms dependence of the ϒ(1S) nuclear modification factor in 2.03 <ycms < 3.05, while they overestimate the results obtained in −4.46 < ycms <−2.96. Furthermore, while the pTdependence of the ALICE measurements indicate slightly stronger cold nuclear matter effects at low pT, the shadowing calculations suggest a flatter behaviour. Finally, the ycms dependence of the RpPb is also compared with a model which includes the effects of parton coherent energy loss with or without the contribution of the EPS09 nuclear shadowing [13,14]. The model predicts a mild dependence of the energy loss mechanism on rapidity. When the nuclear shadowing contribution is included, the model describes the forward-rapidity results, while it slightly overestimates the backward-rapidity RpPb. The ϒ(1S) RpPb is also compared with a theoretical model which includes a shadowing contribution, based on the nCTEQ15 set of nPDFs, on top of a suppression of the ϒ(1S) production due to interactions with comoving particles [15,52]. The uncertainties associated to this theoretical calculation include a small contribution from the uncertainty on the comovers cross section and are dominated by the uncertainties on the shadowing. Also in this case the calculation slightly overestimates the ALICE measurements at backward ycms, while at forward ycms the data agree with the model. It can be noted that the interpretation of the ϒ(1S) behaviour in p–Pb collisions would also benefit from a precise knowledge, so far still affected by large uncertainties, of the feed-down contribution of the excited states into the ϒ(1S). The ϒ(1S) nuclear modification factor is evaluated as a function of the collision centrality. The QpPb results, shown in Fig. 8, are presented as a function of the average number of collisions, Ncoll and it can be observed that both at forward and backward rapidity the ϒ(1S) centrality dependence is rather flat. Finally, the nuclear modification factor is also evaluated for the ϒ(2S) and ϒ(3S) resonances, in the forward and backward-ycms intervals, as shown in Fig. 9. The corresponding ϒ(2S) RpPb values are: Rϒ(2S) pPb (2.03 <ycms <3.53) =0.59 ±0.12 (stat.) ±0.05 (uncor.syst.) ±0.02 (cor.syst.) Rϒ(2S) pPb (−4.46 <ycms <−2.96) =0.69 ±0.12 (stat.) ±0.05 (uncor.syst.) ±0.02 (cor.syst.) the ϒ(2S) suppression being compatible with unity within 3.1σat forward ycms and 2.3σat backward ycms. The ϒ(3S) RpPb values are: Rϒ(3S) pPb (2.03 <ycms <3.53) =0.32 ±0.24 (stat.) ±0.06 (uncor.syst.) ±0.01 (cor.syst.) Rϒ(3S) pPb (−4.46 <ycms <−2.96) =0.71 ±0.23 (stat.) ±0.09 (uncor.syst.) ±0.02 (cor.syst.) The ϒ(3S) suppression is compatible with unity within 2.7σat forward ycms and 1.2σat backward ycms. The difference in the RpPb of the ϒ(2S) and ϒ(1S) amounts to 0.5σin both rapidity intervals, suggesting, in p–Pb collisions, a similar modification of the production yields of the two ϒstates, with respect to pp collisions. Unfortunately, the large uncertainties on the ϒ(3S) prevent robust conclusions on the behaviour of the most loosely bound bottomonium state. The model which includes both the nuclear shadowing contribution (nCTEQ15) and interactions with comoving particles [15,52]suggests a small difference between the nuclear modification factors of the three ϒstates. This difference is slightly more important in the backward-rapidity range, while it becomes negligible at forward ycms. By evaluating the ratio of the ϒ(nS) to ϒ(1S) nuclear modification factors, the shadowing contribution and most of the theory uncertainties, as well as some of the uncertainties on the data, cancel out. The shape of the theoretical calculation is, hence, mainly driven by the interactions with the comoving particles, which affect mostly the excited ϒstates in the backward rapidity region. As shown in the lower panel of Fig. 9, the ALICE measurements and the model are in fair agreement, even if the uncertainties on the data do not yet allow a firm conclusion on the role of comovers to be drawn. 5. Conclusions The ALICE measurements of the rapidity, transverse momentum and centrality dependence of the inclusive ϒ(1S) nuclear modification factor in p–Pb collisions at √sNN =8.16 TeV have been presented. The results show a suppression of the ϒ(1S) yields, with respect to the ones measured in pp collisions at the same centre-of-mass energy. The RpPb values are similar at forward and backward rapidity with a slightly stronger suppression at low pT, while in both rapidity intervals there is no evidence for a centrality dependence of the ϒ(1S) QpPb. The results obtained at √sNN = 8.16 TeV are similar within uncertainties to those measured by ALICE in p–Pb collisions at the lower energy of √sNN =5.02 TeV and show a good agreement with the LHCb measurements at the same centre-of-mass energy. Models based on nuclear shadowing, coherent parton energy loss or interactions with comoving particles fairly describe the data at forward rapidity, while they tend to overestimate the RpPb at backward ycms. The ϒ(2S) RpPb has also been measured, showing a strong suppression, similar to the one measured for the ϒ(1S) in the two investigated rapidity intervals. Finally, a first measurement of the ϒ(3S) has also been performed, even if the large uncertainties prevent a detailed comparison of its behaviour in p–Pb collisions with respect to the other bottomonium states. These new bottomonium measurements represent an 8ALICE Collaboration / Physics Letters B 806 (2020) 135486 Fig. 7. ϒ(1S) RpPb as a function of pTfor Pb–p (left panel) and p–Pb collisions (right panel). The RpPb values are compared with theoretical calculations based on EPS09 NLO [14,48], nCTEQ15 and EPPS16 [9–11,49–51]shadowing implementations. Details on the theory uncertainty bands are discussed in the text. Fig. 8. ϒ(1S) QpPb as a function of Ncoll, for Pb–p (left panel) and p–Pb collisions (right panel). important baseline for the understanding of the role of CNM effects in p–Pb collisions and open up the way for future precision analyses with the upcoming LHC Run 3 and Run 4 data taking periods. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements The ALICE Collaboration would like to thank all its engineers and technicians for their invaluable contributions to the construction of the experiment and the CERN accelerator teams for the outstanding performance of the LHC complex. The ALICE Collaboration gratefully acknowledges the resources and support provided by all Grid centres and the Worldwide LHC Computing Grid (WLCG) collaboration. The ALICE Collaboration acknowledges the following funding agencies for their support in building and running the ALICE detector: A. I. Alikhanyan National Science Laboratory (Yerevan Physics Institute) Foundation (ANSL), State Committee of Science and World Federation of Scientists (WFS), Armenia; Austrian Academy of Sciences, Austrian Science Fund (FWF): [M 2467-N36] and Nationalstiftung für Forschung, Technologie und Entwicklung, Austria; Ministry of Communications and High Technologies, National Nuclear Research Center, Azerbaijan; Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), Financiadora de Estudos e Projetos (Finep), Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) and Universidade Federal do Rio Grande do Sul (UFRGS), Brazil; Ministry of Education of China (MOEC), Ministry of Science & Technology of China (MSTC) and National Natural Science Foundation of China (NSFC), China; Ministry of Science and Education and Croatian Science Foundation, Croatia; Centro de Aplicaciones Tecnológicas y Desarrollo Nuclear (CEADEN), Cubaenergía, Cuba; Ministry of Education, Youth and Sports of the Czech Republic, Czech Republic; The Danish Council for Independent Research | Natural Sciences, the Villum Fonden and Danish National Research Foundation (DNRF), Denmark; Helsinki Institute of Physics (HIP), Finland; Commissariat à l’Énergie Atomique (CEA), Institut National de Physique Nucléaire et de Physique des Particules (IN2P3) and Centre National de la Recherche Scientifique (CNRS) and Région des Pays de la Loire, France; Bundesministerium für Bildung und Forschung (BMBF) and GSI Helmholtzzentrum für Schwerionenforschung GmbH, Germany; General Secretariat for Research and Technology, Ministry of ALICE Collaboration / Physics Letters B 806 (2020) 135486 15 126 University of Jyväskylä, Jyväskylä, Finland 127 University of Liverpool, Liverpool, United Kingdom 128 University of Science and Technology of China, Hefei, China 129 University of South-Eastern Norway, Tonsberg, Norway 130 University of Tennessee, Knoxville, TN, United States 131 University of the Witwatersrand, Johannesburg, South Africa 132 University of Tokyo, Tokyo, Japan 133 University of Tsukuba, Tsukuba, Japan 134 Université Clermont Auvergne, CNRS/IN2P3, LPC, Clermont-Ferrand, France 135 Université de Lyon, Université Lyon 1, CNRS/IN2P3, IPN-Lyon, Villeurbanne, Lyon, France 136 Université de Strasbourg, CNRS, IPHC UMR 7178, F-67000 Strasbourg, France 137 Université Paris-Saclay Centre d’Etudes de Saclay (CEA), IRFU, Départment de Physique Nucléaire (DPhN), Saclay, France 138 Università degli Studi di Foggia, Foggia, Italy 139 Università degli Studi di Pavia, Pavia, Italy 140 Università di Brescia, Brescia, Italy 141 Variable Energy Cyclotron Centre, Homi Bhabha National Institute, Kolkata, India 142 Warsaw University of Technology, Warsaw, Poland 143 Wayne State University, Detroit, MI, United States 144 Westfälische Wilhelms-Universität Münster, Institut für Kernphysik, Münster, Germany 145 Wigner Research Centre for Physics, Budapest, Hungary 146 Yale University, New Haven, CT, United States 147 Yonsei University, Seoul, Republic of Korea iDeceased. ii Dipartimento DET del Politecnico di Torino, Turin, Italy. iii M.V. Lomonosov Moscow State University, D.V. Skobeltsyn Institute of Nuclear, Physics, Moscow, Russia. iv Department of Applied Physics, Aligarh Muslim University, Aligarh, India. vInstitute of Theoretical Physics, University of Wroclaw, Poland.