Elliptic flow of electrons from beauty-hadron decays extracted from Pb–Pb collision data at \(\sqrt{s_\mathrm{NN}} = 2.76\ \hbox {TeV}\)

Eur. Phys. J. C Elliptic flow of electrons from beauty-hadron decays extracted from Pb-Pb collision data at √sNN = 2.76 TeV D. Moreira de Godoy 2 F. Herrmann 2 M. Klasen 2 C. Klein-Bösing 1 2 A. A. P. Suaide 0 0 Universidade de São Paulo , R. do Matão 1371, São Paulo , Brazil 1 ExtreMe Matter Institute EMMI, GSI Helmholtzzentrum für Schwerionenforschung , Planckstraße 1, 64291 Darmstadt , Germany 2 Westfälische Wilhelms-Universität Münster , Wilhelm-Klemm-Straße 9, Münster , Germany We present a calculation of the elliptic flow of electrons from beauty-hadron decays in semi-central PbPb collisions at centre-of-mass energy per colliding nucleon pair, represented as √sNN, of 2.76 TeV. The result is obtained by the subtraction of the charm-quark contribution in the elliptic flow of electrons from heavy-flavour hadron decays in semi-central Pb-Pb collisions at √sNN = 2.76 TeV recently made publicly available by the ALICE collaboration. 1 Introduction The nuclear matter exposed to conditions of high temperature and energy density is expected to undergo a phase transition to a colour deconfined state of matter [ 1,2 ], the Quark-Gluon Plasma (QGP). The conditions for the phase transition can be achieved in the laboratory with collisions of heavy ions at high energies [1]. The properties of the medium formed in the laboratory can be probed with a unique degree of control by particles from decays of heavy flavours (charm and beauty), since heavy quarks are mainly produced in hard parton scattering processes [ 3–5 ] at the initial stage of heavyion collisions [ 6,7 ] and participate in the entire evolution of the created system. The partons traversing the medium lose energy via collisional and radiative processes [ 8–12 ] in the interaction with the medium constituents. The energy loss of partons is predicted to be dependent on their colour charge and mass, resulting in a hierarchy where beauty quarks lose less energy than charm quarks, charm quarks lose less energy than light quarks, and quarks lose less energy than gluons [ 10,13,14 ]. The heavy-flavour energy loss can be investigated experimentally with the nuclear modification factor (RAA) of heavy-flavour particles, defined as (1) RAA = 1 TAA d NAA/d pT dσpp/d pT where d NAA/d pT is the transverse momentum ( pT) differential yield in nucleus-nucleus (AA) collisions; TAA is the average nuclear overlap function in nucleus-nucleus collisions, given by the ratio of the average number of binary collisions and the inelastic cross section; and dσpp/d pT is the pT-differential cross section in proton-proton (pp) collisions. A suppression of the yield of D mesons and leptons from heavy-flavour hadron decays (RAA < 1) for pT > 3 GeV/c was observed in gold–gold (Au–Au) collisions at √sNN = 200 GeV at the Relativistic Heavy Ion Collider (RHIC) [ 15–18 ] and in lead-lead (Pb–Pb) collisions at √sNN = 2.76 and 5.02 TeV at the Large Hadron Collider (LHC) [ 19–25 ], indicating energy loss of heavy flavours in the medium. An experimental hint to the quark mass dependence of the heavy-flavour energy loss has been found in the comparison of the RAA of D mesons and non-prompt J/ψ from B-hadron decays in central Pb–Pb collisions at √sNN = 2.76 TeV at the LHC [ 20,26,27 ]. The observed difference of these measurements is described by model calculations [ 28,29 ] as predominantly due to the quark mass dependence of the parton energy loss. The interaction of heavy quarks with the medium can be further investigated with the azimuthal anisotropy of heavyflavour particles, extracted from the coefficients vn of the Fourier decomposition of the particle azimuthal distribution in the transverse plane [ 30 ] d N d (ϕ − Ψn) ∝ 1 + 2 ∞ n=1 where ϕ is the azimuthal angle of the heavy-flavour particles and Ψn is the symmetry-plane angle of the nth-order harmonic. The second Fourier coefficient v2, called elliptic vn cos [n (ϕ − Ψn)] , (2) flow, quantifies the elliptic azimuthal anisotropy of the emitted particles. The origin of the elliptic azimuthal anisotropy of heavy-flavour particles in non-central heavy-ion collisions depends on the transverse-momentum interval. While the v2 at low pT is sensitive to the collective motion of the medium constituents caused by pressure gradients, the v2 at high pT can constrain the path-length dependence of the in-medium energy loss of heavy quarks, resulting from the direction of the particles that traverse the ellipsoidal nuclear overlap region. The elliptic flow of prompt D mesons at midrapidity is observed to be positive in 30–50% Pb–Pb collisions at √sNN = 2.76 TeV at the LHC [ 31,32 ] with 5.7σ significance in the interval 2 < pT < 6 GeV/c, indicating that charm quarks participate in the collective motion of the system. Measurements of the prompt D-meson v2 in Pb–Pb collisions at √sNN = 5.02 TeV [ 33,34 ] have smaller uncertainties compared to the ones in Pb–Pb collisions at √sNN = 2.76 TeV. The results at the two collision energies are compatible within uncertainties. The prompt Ds+v2 in semi-central Pb–Pb collisions at √sNN = 5.02 TeV is compatible within uncertainties with the average of prompt D0, D+, and D∗+v2 in the same collision system [ 34 ]. A positive v2 is also observed for leptons from heavy-flavour hadron decays at low and intermediate pT in semi-central Au–Au collisions at √sNN = 200 GeV at RHIC [ 17,35 ] and in semi-central Pb–Pb collisions at √sNN = 2.76 TeV at the LHC [ 25,36,37 ]. In particular, the v2 of electrons from heavy-flavour hadron decays is observed to be positive with 5.9σ significance in the range 2 < pT < 2.5 GeV/c in 20– 40% Pb–Pb collisions at √sNN = 2.76 TeV. In view of the experimental results on the elliptic flow of heavy-flavour particles, an important question that remains open is whether beauty quarks take part in the collective motion in the medium. The first measurement of the v2 of non-prompt J/ψ mesons from B-hadron decays is compatible with zero within uncertainties in two kinematic regions, 6.5 < pT < 30 GeV/c and |y| < 2.4, and 3 < pT < 6.5 GeV/c and 1.6 < |y| < 2.4, in 10–60% Pb–Pb collisions at √sNN = 2.76 TeV at the LHC [ 27 ]. In this paper, we present a method to subtract the contribution of charm quarks in the published measurement of the elliptic flow of electrons from heavy-flavour hadron decays in semi-central Pb–Pb collisions at √sNN = 2.76 TeV performed by the ALICE collaboration. The calculation uses as input the v2 coefficients of prompt D mesons and electrons from heavyflavour hadron decays measured by the ALICE collaboration [ 32,36 ] and three different results for the relative contribution of electrons from beauty-hadron decays to the yield of electrons from heavy-flavour hadron decays [ 38–41 ]. 2 Methodology The particle azimuthal distribution of electrons from heavyflavour hadron decays (e ← c + b) can be separated into the contributions of electrons from charm-hadron decays (e ← c) and from beauty-hadron decays (e ← b). Consequently, the elliptic flow of electrons from beauty-hadron decays can be expressed as (3) v2e←b = v2e←c+b − (1 − R)v2e←c R , where R represents the relative contribution of electrons from beauty-hadron decays to the yield of electrons from heavyflavour hadron decays. In the following, we present the currently published measurements and, in case there is no available measurement, our calculations of the three observables required to obtain the elliptic flow of electrons from beauty-hadron decays. Based on available results on open heavy flavours at RHIC and LHC, the most suitable system for this analysis is the Pb–Pb collision system at √sNN = 2.76 TeV in the 20– 40% centrality class, which corresponds to the centrality range where the measured v2 of electrons from heavy-flavour hadron decays is observed to be positive with a maximum significance [ 36 ] and thus a possible elliptic flow of electrons from beauty-hadron decays is expected to be more significant. In this analysis, the v2 and RAA of heavyflavour particles are assumed to be the same at slightly different mid-rapidity ranges (|y| < 0.5, 0.7 and 0.8) in which the measurements needed in the calculation are available. Indeed, no dependence on rapidity was observed in recent ALICE results on those observables for electrons from heavy-flavour hadron decays at mid-rapidity (|y| < 0.7 for v2 and |y| < 0.6 for RAA measurements) and muons from heavy-flavour hadron decays at forward rapidity (2.5 < y < 4) [ 24,36 ]. 2.1 Elliptic flow of electrons from heavy-flavour hadron decays The result on the elliptic flow of electrons from heavy-flavour hadron decays (v2e←c+b) at mid-rapidity (|y| < 0.7) in 20– 40% Pb–Pb collisions at √sNN = 2.76 TeV published by the ALICE collaboration [36] is used in this analysis. The ve←c+b is measured in the interval 0.5 < pT < 13 GeV/c 2 with the event plane method [ 30 ]. A positive value is observed in the interval 2 < pT < 2.5 GeV/c with significance of 5.9σ [ 36 ]. 2.2 Relative contribution of electrons from beauty-hadron decays to the yield of electrons from heavy-flavour hadron decays The measurement of the relative contribution of electrons from beauty-hadron decays to the yield of electrons from heavy-flavour hadron decays (R) has been published by the ALICE collaboration only in pp collisions at √s = 2.76 TeV [ 38,39 ]. The factor R is measured using the track impact parameter and electron-hadron azimuthal correlation methods. Results obtained with both techniques are compatible within uncertainties. The coefficient R measured in pp collisions with the electron-hadron azimuthal correlation method is used in the analysis with the caveat that initial- and final-state effects modify the yield of electrons from heavyflavour hadron decays in heavy-ion collisions. In particular, the coefficient R at high pT is expected to be higher in Pb– Pb collisions compared to pp collisions, since the in-medium energy loss of charm quarks is predicted to be larger than the one of beauty quarks [23]. Therefore, the factor R at high pT in Pb–Pb collisions at √sNN = 2.76 TeV is expected to have an exclusive value between the measured factor R in pp collisions at √s = 2.76 TeV and unity. Consequently, according to Eq. 3, the minimum value of the v2 of electrons from beauty-hadron decays can be computed with the R measured in pp collisions. In addition to the available measurement, the coefficient R is obtained with a Monte Carlo (MC) simulation based on POWHEG [ 42 ], which provides the calculation of the heavyflavour production in hadronic collisions at Next-to-Leading Order (NLO) accuracy. The POWHEG results are interfaced to PYTHIA [ 43,44 ] in order to generate the shower, hadronisation and decay. In agreement with other heavy-flavour production tools, e.g. pQCD calculation at Fixed Order plus Next-to-Leading Logarithms (FONLL) [ 4,45 ] and earlier pp calculations [ 5 ], the square root of the quadratic sum of the quark mass (m Q ) and pT are used as renormalization and factorization scales, i.e. μ = μ f = μr = m2Q + pT2. The charm- and beauty-quark masses are set as 1.5 and 4.75 GeV, respectively. Even though the calculated coefficient R is sensitive to the choice of heavy-quark masses and scales [ 5 ], only the central value is shown in this analysis. Admittedly, the described framework is designed for pp collisions, but by making use of the EPS09 [ 46 ] NLO nuclear Parton Distribution Functions (nPDFs) the framework is able to account for initial-state cold nuclear effects. The nPDF gluon shadowing results in reduced pT-differential cross sections for electrons from heavy-flavour hadron decays for pT < 6 GeV/c and affects contributions from charm quarks stronger than those from beauty quarks. Thus, it provides a lower baseline for the factor R, which is suggested to be further enhanced by medium interactions as it will be discussed in this paper. In e 1 →0.9 b+0.8 /c0.7 e0.6 →)c0.5 addition, the R coefficient is also obtained with a leading order (LO) calculation based on PYTHIA using EPS09 LO nPDFs to study the impact of NLO corrections [ 5 ]. The comparison of the calculations with LO and NLO approaches, shown in Fig. 1, reveals that the factor R is reduced with the NLO corrections, stressing the importance of NLO calculations. In fact, the additional processes of heavy-flavour production at NLO give rise to large logarithmic corrections to the charm- and beauty-quark cross sections depending on the heavy-flavour mass. The corresponding FONLL calculation of the factor R using CTEQ6.6 PDFs, which is also shown in Fig. 1, is similar to the POWHEG+PYTHIA(EPS09NLO) result at high pT. The result on the factor R from the BAMPS heavyflavour transport model [ 40,41 ], which includes collisional and radiative in-medium energy loss of heavy quarks, is also employed in the analysis to obtain the v2 of electrons from beauty-hadron decays. The choice of the BAMPS model is justified by the good agreement of the predictions for the RAA of electrons from beauty- and heavy-flavour hadron decays for pT > 3 GeV/c in central Pb–Pb collisions at √sNN = 2.76 TeV with what measured by the ALICE collaboration [ 23,24 ]. In this analysis, the R coefficient measured in pp collisions using the electron-hadron azimuthal correlation technique and the ones obtained with POWHEG + PYTHIA (EPS09NLO) and with the BAMPS model are used to estimate the elliptic flow of electrons from beauty-hadron decays. 2.3 Elliptic flow of electrons from charm-hadron decays The elliptic flow of electrons from charm-hadron decays (v2e←c) is estimated using a MC simulation of decays of D0 mesons into electrons with PYTHIA. The MC simulation is based on two observables measured for D0 mesons in Pb–Pb collisions at √sNN = 2.76 TeV: – the pT-differential yield, which is used as a probability distribution for finding a D0 meson with a certain pT; – the pT-differential v2, which is used to obtain the ϕD0 − Ψ2 probability distribution with Eq. 2. In fact, the pT-differential yield of D0 mesons at midrapidity (|y| < 0.8) in 20–40% Pb–Pb collisions at √sNN = 2.76 TeV is estimated from the ALICE results on the pTdifferential yield and RAA of prompt D0 mesons at midrapidity (|y| < 0.5) in 0–20% Pb–Pb collisions at the same collision energy [ 19 ] as d NAA d pT , (4) where the coefficient CΔy = 1.6 corresponds to the scaling factor of the yield from |y| < 0.5 to |y| < 0.8 in pp collisions, assuming a uniform distribution of the D0-meson yield within the rapidity range. The terms C TAA = 0.362 ± 0.020 [ 19 ] and C RAA are the ratios of the average nuclear overlap function and the D0-meson RAA, respectively, in Pb–Pb collisions at √sNN = 2.76 TeV in the 20–40% centrality class to the ones in the 0–20% centrality class. Note that the terms CΔy and C TAA are constant, so they do not play a role in the determination of the D0-meson pT probability distribution. The non-measured RAA of D0 mesons in 20–40% Pb–Pb collisions at √sNN = 2.76 TeV is estimated by the average of the ALICE results on the D0-meson RAA in Pb–Pb collisions at the same collision energy in the 0–20 and 40–80% centrality classes [ 19 ] weighted by the corresponding yield of D0 mesons in each centrality class. The resulting pT distribution of D0 mesons in 20–40% Pb–Pb collisions obtained from Eq. 4 is then fitted by a power-law function (top panel of Fig. 2), considering the statistical uncertainty of the experimental results. The fit function is used as the D0-meson pT probability distribution. The v2 of prompt D0 mesons in 20–40% Pb–Pb collisions at √sNN = 2.76 TeV (bottom panel of Fig. 2) is obtained by the arithmetic average of the measured prompt D0-meson v2 in Pb–Pb collisions at the same collision energy in the 10– 30 and 30–50% centrality classes [ 32 ]. Indeed, experimental results show that the v2 of heavy-flavour particles increases with the centrality class [ 17, 32, 36, 37 ], which is consistent with the qualitative expectation of increasing of the elliptic anisotropy from central to peripheral nucleus-nucleus collisions. The statistical and systematic uncertainties are propagated considering the prompt D0-meson v2 in the 10–30 and ) 10 V e /G 1 c ( p1T0−1 d / dN10−2 10−3 10−4 10−5 0 D 0.4 t p rpom 02.3 v 0.2 0.1 0 −0.1 30–50% centrality classes as uncorrelated as a conservative estimation. In the D0-meson v2 measurement by the ALICE collaboration, the central value was obtained by assuming that the v2 coefficients of prompt D mesons and D mesons from B-meson decays are the same [ 32 ]. However, the systematic uncertainty related to this assumption, referred to as systematic uncertainty from the B feed-down subtraction, was evaluated by the ALICE collaboration. It was assumed that the v2 of prompt D mesons from B-meson decays should be between zero and v2 of prompt D mesons, resulting in the upper and lower limits of the systematic uncertainty, respectively. Therefore, the B feed-down contribution decreases the absolute value of the D0-meson v2 and thus the systematic uncertainty is restricted to the upper (lower) limit when the v2 is positive (negative). Since the measured D0-meson v2 coefficients are negative in the 8 < pT < 12 and 12 < pT < 16 GeV/c intervals in the 10–30 and 30–50% centrality classes, respectively, the resulting propagated systematic uncertainty from the B feed-down subtraction contains lower and upper limits. Finally, the estimated pT and v2 distributions of D0 mesons in 20–40% Pb–Pb collisions at √sNN = 2.76 TeV are used to obtain the v2e←c = cos [2 (ϕe − Ψ2)] in the same collision system using the PYTHIA event generator. The azimuthal angle of electrons (ϕe) takes into account the angular separation between electrons and their parent D0 mesons. 2.3.1 Statistical uncertainty The statistical uncertainty of the D0-meson v2 is used as input for the MC simulation to obtain the statistical uncertainty of the ve←c. The statistical uncertainties of the measurements 2 used to obtain the pT-differential yield of D0 mesons are considered in the fit of the D0-meson probability distribution. Further variations are considered as systematic uncertainties. 2.3.2 Systematic uncertainty The systematic uncertainties from data and from the B feeddown subtraction of the D0-meson v2 (bottom panel of Fig. 2) are used as input for the MC simulation to obtain the systematic uncertainty of the ve←c. The following is a discussion 2 on other sources of systematic uncertainty that can influence the v2e←c estimation. In order to validate the Eq. 4, the pT-differential yield and RAA of prompt D0 mesons in 40–80% Pb–Pb collisions at √sNN = 2.76 TeV [ 19 ] are also used as reference to obtain the pT-differential yield of D0 mesons in the 20–40% centrality class. The result is the same as the one obtained with the 0–20% centrality class (top panel of Fig. 2). The ALICE result on the D0-meson RAA in the 30–50% centrality class [ 32 ] is used as an alternative for the D0-meson RAA estimation in the 20–40% centrality class. No significant difference is observed in the resulting ve←c with respect to 2 the one obtained with the RAA estimated by the average of the D0-meson RAA measurements in the 0–20 and 40–80% centrality classes weighted by the corresponding yield of D0 mesons in each centrality class. The systematic uncertainties of the measurements of the pT-differential yield and RAA of prompt D0 mesons are considered in the fit of the D0-meson pT distribution in 20–40% Pb–Pb collisions. No significant difference is observed in the resulting ve←c with respect to the one considering only the 2 statistical uncertainty in the fit. For further investigation, The BAMPS result on the pT distribution of D mesons at |y| < 0.8 in 20–40% Pb–Pb collisions at √sNN = 2.76 TeV [ 40,41 ] is also used to compute the v2e←c. The relative difference of the obtained v2e←c using the estimated pT distribution of D0 mesons and the BAMPS result, which increases from 1 to 20% in the interval 2 < pT < 8 GeV/c, is included in the systematic uncertainty. The effect of the D0-meson v2 estimation in 20–40% Pb–Pb collisions using the arithmetic average of the D0meson v2 measurements in the 10–30 and 30–50% centrality classes is investigated in this analysis. For this purpose, the trend of the unidentified charged particle v2 as a function of the average number of binary collisions ( Ncoll ) [ 47 ] is assumed to be the same as the one for D0 mesons. The v2 as a function of Ncoll is obtained from a parametrisation of the centrality-dependent v2 measurement of unidentified charged particles integrated over the interval 0.2 < pT < 5GeV/c [ 48 ]. The corresponding result exhibits a linear dependence between v2 and Ncoll with a negative slope for Ncoll > 220. For comparison, the parametrisation is also obtained from the centrality-dependent v2 measurement of unidentified charged particles integrated over the interval 10 < pT < 20 GeV/c [ 49 ]. The linear dependence between v2 and Ncoll is the same as the one obtained for particles in a lower pT interval. The D0-meson v2 is then obtained by the average of the D0-meson v2 in the 10–30 and 30–50% centrality classes weighted by the v2 coefficients of the corresponding Ncoll values [ 47 ]. The relative difference of the obtained D0-meson v2 with respect to the one obtained with the arithmetic average is negligible for pT < 8 GeV/c and its average is 19% for pT > 8 GeV/c, which is still compatible within uncertainties. The v2e←c coefficients obtained with the two approaches show a relative difference of 2% in the range 2 < pT < 8 GeV/c. This deviation is considered as a consequence of statistical fluctuations in the D0-meson v2 measurement for pT > 8 GeV/c and thus no systematic uncertainty is assigned for this effect. In order to investigate the impact of the assumption of the particle mass ordering of the elliptic flow [ 50 ] used to determine the systematic uncertainty from the B feed-down subtraction in the D0-meson v2 measurement, one can assume that the v2 of prompt D mesons from B-meson decays should be between zero and the unidentified charged particle v2. The unidentified charged particle v2 in 20–40% Pb–Pb collisions is obtained by the average of the v2 measurements in the 20–30 and 30–40% centrality classes [ 49 ] weighted by the corresponding Ncoll values. The v2 coefficients of prompt D0 mesons and unidentified charged particles are compatible within uncertainties as well as the ve←c obtained with 2 these two results. Therefore, the lower limit of the systematic uncertainty from the B feed-down subtraction can be positioned at the central values of the prompt D-meson v2 and v2e←c without strictly considering that the B-meson v2 is expected to be lower than the D-meson v2. As a consequence of the pT interval (2 < pT < 16 GeV/c) of the D0-meson v2 and pT-differential yield measurements, the v2e←c is obtained in the range 2 < pT < 8 GeV/c. The fraction of electrons with pT > 2 GeV/c that come from D0 mesons with pT < 2 GeV/c is negligible according to PYTHIA simulations. The effect of the pT upper limit of the D0-meson measurements is studied by evaluating the ve←c with extrapolation of the pT and v2 2 distributions of D0 mesons up to 26 GeV/c. The transverse momentum extrapolation is obtained from the power-law fit function shown in the top panel of Fig. 2, while the impact of the v2 of D0 mesons is estimated by explicitly setting its value, in the interval 16 < pT < 26 GeV/c, to either zero, or constant at high pT, or maximum value of the prompt D0meson v2 (shown in Fig. 2). The highest relative difference in these three scenarios, which increases from 0.3 to 40% in the interval 2 < pT < 8 GeV/c, is assigned as a conservative systematic uncertainty. The effect of the mid-rapidity range of D0 mesons is investigated by obtaining the ve←c using the D0-meson pT dis2 tribution in the rapidity range |y| < 1.6 as input for the simulation. The D0-meson v2 is considered to be the same in this rapidity range, because no dependence on rapidity was observed in ALICE results on leptons from heavy-flavour hadron decays [ 24, 36 ] as discussed previously. The relative difference of the obtained v2e←c with respect to the one using the D0-meson pT distribution in the rapidity range |y| < 0.8 is negligible and thus no additional systematic uncertainty is considered due to the rapidity effect. In this analysis, the v2 and shape of the pT-differential yields of charm hadrons are assumed to be the same as the ones measured for D0 mesons. This is justified by the fact that the v2 coefficients of D0, D+ and D∗+ mesons are compatible within uncertainties in 30–50% Pb–Pb collisions at √sNN = 2.76 TeV [ 31 ], also the prompt Ds+v2 is compatible within uncertainties with the prompt non-strange D meson v2 in 30–50% Pb–Pb collisions at √sNN = 5.02 TeV [ 34 ]. In addition, the ratios of the yields of D+/D0 and D∗+/D0 were observed to be constant within uncertainties in pp collisions at √s = 7 TeV and no modification of the ratios was observed within uncertainties in central and semi-central Pb– Pb collisions at √sNN = 2.76 TeV [ 51 ]. A possible hint for an enhancement of the Ds+/D0 ratio is observed in 0–10% Pb–Pb collisions at √sNN = 2.76 TeV [ 52 ], but the current uncertainties do not allow for a conclusion. The effect of different decay kinematics of charm particles is estimated by simulating the v2 of electrons from combined D, D∗, Ds , and Λc particle decays taking into account the fraction of charm quarks that hadronise into these particles [ 53 ] and using the same simulation input as used in the analysis ( pTdifferential yield and v2 of D0 mesons). The obtained ve←c 2 is compatible with the one using D0-meson decay and thus no systematic uncertainty is considered due to this effect. In order to exemplify the impact of a possible production enhancement of Ds+ and Λc particles in Pb–Pb collisions with respect to pp collisions, their fragmentation fractions are increased by a factor 2 and 5, respectively, in the simulation of the combined charm meson v2. The relative difference of the obtained ve←c and the one using D0-meson 2 decay is negligible for pT < 3 GeV/c and its average is 5% for pT > 3 GeV/c. Finally, the D0-meson v2 systematic uncertainty from data is summed in quadrature with other sources of systematic uncertainty that affect significantly the v2e←c estimation, which are the pT distribution of D0 mesons and the limited pT interval of the D0-meson measurements. They are considered as uncorrelated since the effect from the pT distribution of D0 mesons is obtained with the BAMPS result and the effect from the limited pT interval of the D0-meson measurements is obtained by extrapolations. The term “from data” is maintained later in this paper to distinguish all sources of systematic uncertainty from the systematic uncertainty related to the B feed-down subtraction of the D0-meson v2 measurement, which is shown separately. 2.4 Elliptic flow of electrons from beauty-hadron decays The v2 of electrons from beauty-hadron decays (v2e←b) is obtained from Eq. 3 using the R, v2e←c+b and ve←c results 2 presented in their respective sections. The three results are considered as statistically independent. First, the factor R was measured in a different collision system (pp collisions) or obtained with calculations. Second, the v2e←c is obtained with a simulation using measurements of D0 mesons reconstructed via the hadronic decay channel D0 → K−π + in a different centrality class than in the ve←c+b measurement. 2 Even though the systematic uncertainties of the R, v2e←c+b and ve←c results might be partially correlated, especially 2 concerning the particle identification selection criteria, the limited public information prevents a more accurate treatment of these uncertainties. Therefore, they are assumed to be uncorrelated as a conservative estimation. As an example of the effect of a possible overestimation, if the systematic uncertainties of the ve←c and ve←c+b results decrease by 2 2 30% in the interval 2 < pT < 8 GeV/c, the systematic uncertainty from data of the ve←b result is expected to decrease 2 by approximately 24%. Therefore, the statistical and systematic uncertainties of the R, ve←c+b and ve←c results are propagated as inde2 2 pendent variables. The ve←b systematic uncertainties from 2 data and from the B feed-down subtraction are asymmetric as a consequence of the systematic uncertainty asymmetry of the measurements used in this analysis. The systematic uncertainty from data is evaluated according to the method described in [ 54 ], where the positive and negative deviations are obtained separately and their average is added in quadrature. For verification, the alternative approach presented in [ 55 ] is also applied in this analysis. No significant difference between these methods is observed. Since the asymmetry of the systematic uncertainty from the B feeddown subtraction only comes from the v2e←c result, the limits of the v2e←b systematic uncertainty are the deviations resulting from the upper and lower limits of the v2e←c systematic uncertainty. 3 Results The relative contribution of electrons from beauty-hadron decays to the yield of electrons from heavy-flavour hadron decays at √s = 2.76 TeV obtained with POWHEG + PYTHIA at NLO accuracy using EPS09 NLO nPDFs is shown in the top panel of Fig. 3. The result is compared with the R in pp collisions at √s = 2.76 TeV measured by the ALICE collaboration using the electron-hadron azimuthal correlation technique [ 38,39 ] and with the BAMPS result in 20–40% Pb–Pb collisions at √sNN = 2.76 TeV [40]. The comparison shows that R is higher when in-medium effects are present, which is consistent with the expectation of the mass hierarchy of the energy loss of charm and beauty quarks in the medium. The bottom panel of Fig. 3 shows the v2 of electrons from charm-hadron decays at mid-rapidity in 20–40% Pb–Pb collisions at √sNN = 2.76 TeV obtained with a MC simulation with PYTHIA using as input the pTdifferential yield and v2 distributions of D0 mesons in the same collision system. A positive v2 of electrons from charmhadron decays is found in all pT intervals, with a maximum significance of 3.2σ−, where σ− is the combined statistical and systematic uncertainties of the lower limit, in the interval 2 < pT < 3 GeV/c. The v2 coefficients of electrons from beauty-hadron decays in 20–40% Pb–Pb collisions at √sNN = 2.76 TeV obtained with different approaches of the factor R (top panel of Fig. 3) are shown in Fig. 4. The result computed with the coefficient R in pp collisions is an estimation of the minimum value, as discussed previously. The v2 of electrons from beauty-hadron decays in 20–40% Pb–Pb collisions at √sNN = 2.76 TeV is compatible with zero within approximately 1σ of the total uncertainty, obtained by summing in quadrature the different uncertainty contributions, in all pT intervals and different R coefficients. However, the large statistical and systematic uncertainties prevent a definite conclusion. The result is consistent with the measured v2 of nonprompt J/ψ mesons from B-hadron decays in 10–60% Pb–Pb collisions at √sNN = 2.76 TeV [ 27 ], which is also compatible with zero within uncertainties. Figure 5 shows the v2 of electrons from charm- and beauty-hadron decays, inclusive [ 36 ] and separated, in 20– 40% Pb–Pb collisions at √sNN = 2.76 TeV. The v2 of electrons from beauty-hadron decays is lower than the v2 of electrons from charm-hadron decays, although they are compatible within uncertainties. The average of the v2 coefficients of electrons from charm- and beauty-hadron decays obtained in the interval 2 < pT < 8 GeV/c are listed in Table 1. Because of the asymmetric uncertainties, the average is obtained numerically with an iterative sum of the likelihood functions parametrised by variable-width Gaussians [ 55,56 ]. The standard deviation, which is the combination of statistical and systematic uncertainties, is assumed to vary linearly. The maximum value of the summed likelihood function corresponds to the average v2, while the points at which the function is −0.5 correspond to the lower and upper limits of the total uncertainty. R from ALICE R from POWHEG+PYTHIA(EPS09NLO) R from BAMPS Fig. 5 Elliptic flow of electrons from charm- and beauty-hadron decays, inclusive [36] and separated, in 20–40% Pb–Pb collisions at √sNN = 2.76 TeV. The vertical error bars represent the statistical uncertainties and the horizontal error bars indicate the bin widths. The empty and filled boxes represent the systematic uncertainties from data and from the B feed-down subtraction, respectively, in the D0-meson v2 measurement [ 32 ] 4 Conclusions We presented a method to subtract the contribution of charm quarks in the elliptic flow of electrons from heavy-flavour hadron decays. The v2 of electrons from charm-hadron decays was estimated using a MC simulation of D0-meson decays into electrons with PYTHIA, based on measurements of the pT-differential yield and v2 of D0 mesons in Pb– Pb collisions at √sNN = 2.76 TeV by ALICE. A positive v2 of electrons from charm-hadron decays is found with a maximum significance of 3.2σ− in the interval 2 < pT < 3 GeV/c. The computed v2 of electrons from charm-hadron decays was then subtracted from the v2 of electrons from ←b0.25 e 2 v 0.2 izontal error bars indicate the bin widths. The empty and filled boxes represent the systematic uncertainties from data and from the B feeddown subtraction, respectively, in the D0-meson v2 measurement [ 32 ] Table 1 Average of the v2 coefficients of electrons from charm- and beauty-hadron decays obtained in the transverse momentum interval 2 < pT < 8 GeV/c in 20–40% Pb–Pb collisions at √sNN = 2.76 TeV. The reported errors are the combined statistical and systematic uncertainties. See text for more details Result e ← c e ← b e ← b e ← b R approach – ALICE BAMPS POWHEG+PYTHIA(EPS09NLO) Average v2 0.150+0.034 −0.028 0.014+0.039 −0.042 −0.010+0.047 −0.052 0.032+0.028 −0.030 heavy-flavour hadron decays in 20–40% Pb–Pb collisions at √sNN = 2.76 TeV measured by the ALICE collaboration. The subtraction was weighted by the relative contribution of electrons from beauty-hadron decays to the yield of electrons from heavy-flavour hadron decays. Since this observable is not measured in Pb–Pb collisions, three different approaches were used as estimations in the analysis. The resulting v2 of electrons from beauty-hadron decays in 20–40% Pb–Pb collisions at √sNN = 2.76 TeV from the subtraction is compatible with zero within approximately 1σ of the total uncertainty in all pT intervals and different approaches of the relative contribution of electrons from beauty-hadron decays to the yield of electrons from heavy-flavour hadron decays. However, the large statistical and systematic uncertainties prevent a definite conclusion. The v2 of electrons from beauty-hadron decays is found to be lower than the v2 of electrons from charm-hadron decays. 5 Outlook In the presented method, the elliptic flow of electrons from beauty-hadron decays can be determined by using three observables that have largely been measured at the LHC and RHIC. Based on available results of these observables, the procedure was applied using measurements performed by the ALICE collaboration. The method demonstrated to be effective; however, the current statistical and systematic uncertainties of the ALICE results prevent a definite conclusion whether the collective motion of the medium constituents influences beauty quarks. A better accuracy of the results on heavy-flavour particles has been achieved in measurements in Pb–Pb collisions at √sNN = 5.02 TeV [ 34 ] and it is expected to be further improved with the ALICE upgrade, which is foreseen to start in 2019. In particular, the upgrade of the Inner Tracking System (ITS) detector will improve the determination of the distance of closest approach to the primary vertex, momentum resolution and readout rate capabilities [ 57 ]. These improvements will allow for more precise measurements of D mesons down to low transverse momenta and for reducing the systematic uncertainties from data and from the B feed-down subtraction. The latter will be possible with the direct measurement of the fraction of prompt D mesons and D mesons from Bmeson decays, which is expected to be accessible with relative statistical and systematic uncertainties smaller than 1 and 5% [ 57 ], respectively, for prompt D0 mesons. In addition, the ITS upgrade will enable the tracking of electrons down to approximately 0.05 GeV/c and enhance the capability to separate prompt from displaced electrons [ 57 ], improving the reconstruction of electrons that do not originate from heavy-flavour hadron decays needed for the background subtraction. Moreover, the systematic uncertainty of the elliptic flow of electrons from beauty-hadron decays can be further improved by taking into account correlations among different contributions. The capability of the heavy-flavour measurements will also enhance with the increase of luminosity. For instance, the current relative statistical uncertainty of the D-meson v2 measurement in Pb–Pb collisions is 10% for an integrated luminosity of 0.1 nb−1, while it is expected to be 0.2% for a scenario with an integrated luminosity of 10 nb−1 [ 57 ]. Also the elliptic flow coefficients of Ds and Λc particles are expected to be achievable with a relative statistical uncertainty of 8 and 20% [ 57 ], respectively. Therefore, the presented method can be used to extract the elliptic flow of electrons from beauty-hadron decays with better precision with future measurements of the three needed observables. Acknowledgements We would like to thank Carsten Greiner and Florian Senzel for providing the BAMPS results, as well as Francesco Prino for fruitful discussions. We are grateful for the support of the Deutsche Forschungsgemeinschaft (DFG) through the Research Training Group “GRK 2149: Strong and Weak Interactions—from Hadrons to Dark Matter”; Bundesministerium für Bildung und Forschung (BMBF) under the project number 05P15PMCA1; Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq); and Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP). Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecomm ons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Funded by SCOAP3. 1. K. Frithjof , Lattice QCD at High Temperature and Density (Springer, Berlin, 2002 ), pp. 209 - 249 . https://doi.org/10.1007/ 3-540-45792- 5 _ 6 2. P. Petreczky, PoS ConfinementX, 028 ( 2012 ). https://pos.sissa.it/ 171/028/pdf 3. E. Norrbin , T. Sjostrand, Eur. Phys. J. C 17 , 137 ( 2000 ). https:// doi.org/10.1007/s100520000460 4. M. Cacciari , S. Frixione , N. Houdeau , M.L. Mangano , P. Nason , G. Ridolfi, J. High Energy Phys . 10 , 137 ( 2012 ). https://doi.org/ 10.1007/JHEP10( 2012 ) 137 5. M. Klasen , C. Klein-Bösing , K. Kovarik , G. Kramer, M. Topp et al., J. High Energy Phys . 1408 , 109 ( 2014 ). https://doi.org/10.1007/ JHEP08( 2014 ) 109 6. B.W. Zhang , C.M. Ko , W. Liu, Phys. Rev. C 77 , 024901 ( 2008 ). https://doi.org/10.1103/PhysRevC.77.024901 7. P. Braun-Munzinger , J. Phys. G 34 ( 8 ), S471 ( 2007 ). http://stacks. iop.org/0954-3899/34/i=8/a=S36 8. E. Braaten , M.H. Thoma , Phys. Rev. D 44 , R2625 ( 1991 ). https:// doi.org/10.1103/PhysRevD.44.R2625 9. M. Djordjevic , Phys. Rev. C 74 , 064907 ( 2006 ). https://doi.org/10. 1103/PhysRevC.74.064907 10. S. Wicks , W. Horowitz , M. Djordjevic , M. Gyulassy , Nucl. Phys. A 783 , 493 ( 2007 ). https://doi.org/10.1016/j.nuclphysa. 2006 . 11 .102 11. R. Baier , Y.L. Dokshitzer , A.H. Mueller , S. Peigne , D. Schiff , Nucl. Phys. B 483 , 291 ( 1997 ). https://doi.org/10.1016/ S0550- 3213 ( 96 ) 00553 - 6 12. M. Djordjevic , U. Heinz, Phys. Rev. C 77 , 024905 ( 2008 ). https:// doi.org/10.1103/PhysRevC.77.024905 13. Y.L. Dokshitzer , D. Kharzeev , Phys. Lett. B 519 , 199 ( 2001 ). https://doi.org/10.1016/S0370- 2693 ( 01 ) 01130 - 3 14. N. Armesto , C.A. Salgado , U.A. Wiedemann, Phys. Rev. D 69 , 114003 ( 2004 ). https://doi.org/10.1103/PhysRevD.69.114003 15. B.I. Abelev et al., Phys. Rev. Lett . 98 , 192301 ( 2007 ). https://doi. org/10.1103/PhysRevLett.98.192301 16. L. Adamczyk et al., Phys. Rev. Lett . 113 , 142301 ( 2014 ). https:// doi.org/10.1103/PhysRevLett.113.142301 17. A. Adare et al., Phys. Rev. C 84 , 044905 ( 2011 ). https://doi.org/ 10.1103/PhysRevC.84.044905 18. S.S. Adler et al., Phys. Rev. Lett . 96 , 032301 ( 2006 ). https://doi. org/10.1103/PhysRevLett.96.032301 19. B. Abelev et al., J. High Energy Phys . 09 , 112 ( 2012 ). https://doi. org/10.1007/JHEP09( 2012 ) 112 20. J. Adam et al., J. High Energy Phys . 11 , 205 ( 2015 ). https://doi.org/ 10.1007/JHEP11( 2015 ) 205 . [Addendum: JHEP 06 , 032 ( 2017 )] 21. B. Abelev et al., Phys. Rev. Lett . 109 , 112301 ( 2012 ). https://doi. org/10.1103/PhysRevLett.109.112301 22. A.M. Sirunyan , et al., CMS-HIN-16-001 , CERN-EP- 2017- 186 ( 2017 ). arXiv: 1708 . 04962 23. J. Adam et al., J. High Energy Phys . 2017 ( 7 ), 52 ( 2017 ). https:// doi.org/10.1007/JHEP07( 2017 ) 052 24. J. Adam et al., Phys. Lett. B 771 , 467 ( 2017 ). https://doi.org/10. 1016/j.physletb. 2017 . 05 .060 25. ATLAS , Tech. Rep. ATLAS-CONF-2015-053 , CERN ( 2015 ). https://cds.cern.ch/record/2055674 26. S. Chatrchyan et al., J. High Energy Phys . 05 , 063 ( 2012 ). https:// doi.org/10.1007/JHEP05( 2012 ) 063 . (CMS-HIN- 10-006, CERNPH-EP- 2011-170) 27. V. Khachatryan et al., Eur. Phys. J. C 77 ( 4 ), 252 ( 2017 ). https:// doi.org/10.1140/epjc/s10052-017-4781-1 28. M. Djordjevic , M. Djordjevic , B. Blagojevic , Phys. Lett. B 737 , 298 ( 2014 ). https://doi.org/10.1016/j.physletb. 2014 . 08 .063 29. A. Andronic et al., Eur. Phys. J. C 76 ( 3 ), 107 ( 2016 ). https://doi. org/10.1140/epjc/s10052-015-3819-5 30. A.M. Poskanzer , S.A. Voloshin , Phys. Rev. C 58 , 1671 ( 1998 ). https://doi.org/10.1103/PhysRevC.58.1671 31. B. Abelev et al., Phys. Rev. Lett . 111 , 102301 ( 2013 ). https://doi. org/10.1103/PhysRevLett.111.102301 32. B.B. Abelev et al., Phys. Rev. C 90 ( 3 ), 034904 ( 2014 ). https://doi. org/10.1103/PhysRevC.90.034904 33. A.M. Sirunyan , et al., CMS-HIN-16-007 , CERN-EP- 2017- 174 ( 2017 ). arXiv: 1708 . 03497 34. S. Acharya et al., Phys. Rev. Lett . 120 , 102301 ( 2018 ). https://doi. org/10.1103/PhysRevLett.120.102301 35. L. Adamczyk et al., Phys. Rev. C 95 , 034907 ( 2017 ). https://doi. org/10.1103/PhysRevC.95.034907 36. J. Adam et al., J. High Energy Phys . 2016 ( 9 ), 28 ( 2016 ). https:// doi.org/10.1007/JHEP09( 2016 ) 028 37. J. Adam et al., Phys. Lett. B 753 , 41 ( 2016 ). https://doi.org/10. 1016/j.physletb. 2015 . 11 .059 38. B.B. Abelev et al., Phys. Lett. B 738 , 97 ( 2014 ). https://doi.org/10. 1016/j.physletb. 2014 . 09 .026 39. B. Abelev et al., Phys. Lett. B 763 , 507 ( 2016 ). https://doi.org/10. 1016/j.physletb. 2016 . 10 .004 40. J. Uphoff , O. Fochler , Z. Xu , C. Greiner , J. Phys . G 42 ( 11 ), 115106 ( 2015 ). https://doi.org/10.1088/ 0954 -3899/42/11/115106. http:// stacks.iop.org/0954-3899/42/i=11/a=115106 41. J. Uphoff , O. Fochler , Z. Xu , C. Greiner , Phys. Lett. B 717 , 430 ( 2012 ). https://doi.org/10.1016/j.physletb. 2012 . 09 .069 42. S. Frixione, P. Nason , G. Ridolfi, J. High Energy Phys . 09 , 126 ( 2007 ). https://doi.org/10.1088/ 1126 - 6708 / 2007 /09/126 43. T. Sjostrand , S. Mrenna , P.Z. Skands , J. High Energy Phys . 05 , 026 ( 2006 ). https://doi.org/10.1088/ 1126 - 6708 / 2006 /05/026. http://stacks.iop.org/1126- 6708 / 2006 /i=05/a=026 44. T. Sjostrand , S. Mrenna , P.Z. Skands , Comput. Phys. Commun . 178 , 852 ( 2008 ). https://doi.org/10.1016/j.cpc. 2008 . 01 .036 45. M. Cacciari , M. Greco , P. Nason , J. High Energy Phys . 1998 ( 05 ), 007 ( 1998 ). http://stacks.iop.org/1126- 6708 / 1998 /i=05/a=007 46. K. Eskola , H. Paukkunen , C. Salgado , J. High Energy Phys . 2009 ( 04 ), 065 ( 2009 ). http://stacks.iop.org/1126- 6708 / 2009 /i=04/ a=065 47. B. Abelev et al., Phys. Rev. C 88 , 0044909 ( 2013 ). https://doi.org/ 10.1103/PhysRevC.88.044909 48. K. Aamodt et al., Phys. Rev. Lett . 105 , 252302 ( 2010 ). https://doi. org/10.1103/PhysRevLett.105.252302 49. B. Abelev et al., Phys. Lett. B 719 , 18 ( 2013 ). https://doi.org/10. 1016/j.physletb. 2012 . 12 .066 50. P.F. Kolb , U.W. Heinz, arXiv:nucl-th/0305084 ( 2003 ) 51. J. Adam et al., J. High Energy Phys . 2016 ( 3 ), 81 ( 2016 ). https:// doi.org/10.1007/JHEP03( 2016 ) 081 52. J. Adam et al., J. High Energy Phys . 2016 ( 3 ), 82 ( 2016 ). https:// doi.org/10.1007/JHEP03( 2016 ) 082 53. C. Amsler et al., Review of Particle Physics. Phys. Lett. B 667 ( 1 ), 1 ( 2008 ). https://doi.org/10.1016/j.physletb. 2008 . 07 .018. http:// www.sciencedirect.com/science/article/pii/S0370269308008435 54. G. D'Agostini , ( 2004 ). arXiv:physics/0403086 55. R. Barlow, ( 2003 ). arXiv:physics/0306138 56. R. Barlow, ( 2004 ). arXiv:physics/0406120 57. B. Abelev et al., J. Phys. G 41 ( 8 ), 087002 ( 2014 ). https://doi.org/10. 1088/ 0954 -3899/41/8/087002. http://stacks.iop.org/0954-3899/ 41/i=8/a=087002