Arxiv:1206.5528V2 [Nucl-Ex] 7 Nov 2012 .J .Codrington, M

arXiv:1206.5528v2 [nucl-ex] 7 Nov 2012 .J .Codrington, M. J. M. .Hirsch, A. .Geurts, F. .R Sahoo, R. N. .R Debbe, R. R. .Ogawa, A. .Mohammed, Y. .A Lisa, A. M. .M Poskanzer, M. A. .Wang, Q. .Markert, C. .vnNieuwenhuizen, van G. nlsv hre arnelpi o nA ucliin a collisions Au + Au in flow elliptic hadron charged Inclusive .H Xu, H. Q. .Djawotho, P. .Huo, L. .Alford, J. .C Cervantes, C. M. .P Sichtermann, P. E. .Kauder, K. .G deSouza, G. U. .G Ma, G. Y. .Banerjee, A. .Pachr, M. .Putschke, J. .G Efimov, G. L. .P Zhang, P. Z. .Pei, H. .G Ritter, G. H. .Adamczyk, L. .P Scharenberg, P. R. .Sat eToledo, de Szanto A. .K Bhati, K. A. .Fedorisin, J. .H Thomas, H. J. .Tribedy, P. .Kisel, I. .Haag, B. .Seyboth, P. .Leight, W. .Bouchet, J. .Butterworth, J. .W Wissink, W. S. .A .Lamont, C. A. M. .Vokal, S. .K Nandi, K. B. .Zawisza, M. .Koroleva, L. .Sumbera, M. .R Stevens, R. J. .Chen, L. 46 37 40 37 10 4 22 43 .Igo, G. 31 .W Hoffmann, W. G. 25 47 44 .Oh, K. .Gibson, A. 52 .L Wang, L. X. 13 .Perkins, C. 4 10 6 .D Anderson, D. B. .Xu, W. 46 21 .Liu, F. .Masui, H. .Kisiel, A. .G Dedovich, G. T. .M .D aaoaetg Don, Madagodagettige D. M. M. D. .Sakrejda, I. 52 .Hajkova, O. 26 55 .S Page, S. B. 52 46 42 .W Ke, W. H. 22 27 34 21 9 25 .Dong, X. .A Voloshin, A. S. 21 1 25 53 .Barnovska, Z. 41 .J LeVine, J. M. .Qiu, H. .Cheng, J. .A Trzeciak, A. B. 17 .Mohanty, B. 25 7 .Zhao, F. 46 .V Brandin, V. A. .Shah, N. .Bichsel, H. .G Fersch, G. R. 19 38 .B Roberts, B. J. .Agakishiev, G. .Elnimr, M. 18 .W Jacobs, W. W. .B Powell, B. C. 40 14 .Zbroszczyk, H. 25 .M Spinka, M. H. 46 d Nasim, Md. 7 .Tian, J. 9 .Chaloupka, P. 18 .Ohlson, A. .Korsch, W. 37 .Witt, R. .Xu, Y. 51 .Z Cai, Z. X. .Ljubicic, T. .M Sun, M. X. .N Singaraju, N. R. .Corliss, R. 4 26 5 .Stock, R. 53 41 25 42 .M Schmah, M. A. .M Landgraf, M. J. 25 .Gliske, S. 25 .Peryt, W. .A afse,Jr., Vanfossen, A. J. 9 25 18 .Kizka, V. 7 13 .Takahashi, J. .S Matis, S. H. 7 .Wang, Y. 49 .Keane, D. .Raniwala, R. .L Drachenberg, L. J. .Zhao, J. 44 .Salur, S. .K Pal, K. S. .Shahaliev, E. 47 42 54 22 .Hamed, A. .Cherney, M. 55 21 30 50 14 .R Timmins, R. A. 4 55 .J Hofman, J. D. 23 .Xu, Z. .Bielcik, J. .D Anson, D. C. 56 52 44 23 40 .Engelage, J. 53 25 15 .M Mondal, M. M. .Deng, J. .Li, C. 29 26 21 .Witzke, W. .R Beavis, R. D. 25 2 .Vossen, A. 4 2 .Filip, P. 18 .Okorokov, V. 14 53 .K Nayak, K. T. .Caines, H. 53 .Kotchenda, L. .V Rogachevskiy, V. O. .Prindle, D. .D Tsai, D. O. .Srivastava, B. .Strikhanov, M. .G Cramer, G. J. 21 .J Llope, J. W. .G Brovko, G. S. .M Aggarwal, M. M. .N Gorbunov, N. Y. 49 44 .Sun, Y. .Jena, C. 25 22 .Chang, Z. .B Zhang, B. J. .Pile, P. 4 52 .R Klein, R. S. 25 42 52 .Webb, G. .Zhong, C. 25 .Sandacz, A. 8 .Xue, L. 46 .Kechechyan, A. 4 39 .X Pan, X. Y. 43 .McDonald, D. .Li, L. .H Tang, H. A. .J Skoby, J. M. 13 21 .Schmidke, B. .LaPointe, S. 12 21 -.Han, L-X. .Raniwala, S. .Draid Souza, de Derradi R. 42 18 .Bielcikova, J. .Shao, M. 23 5 31 .Chikanian, A. .Finch, E. 16 10 56 22 4 54 46 .Eppley, G. 7 .Sun, Z. 48 4 47 .Wada, M. 44 46 .F Wu, F. Y. .Arkhipkin, D. .Planinic, M. 29 .Joseph, J. 52 46 .Horvat, S. .Cleńd aBraSánchez, Barca la Calderón de M. 40 .Varma, R. 23 .Turnau, J. .Bellwied, R. 44 .Pruneau, C. .E Draper, E. J. 37 54 29 6 .Tlusty, D. .Li, W. 9 .Chattopadhyay, S. .Yang, Y. 25 12 .W Oldag, W. E. 29 .M Nelson, M. J. .Morozov, B. .S Longacre, S. R. .C Webb, C. J. .Bruna, E. 53 .Zhu, X. .Zhang, S. 7 .J Crawford, J. H. .D .Stanislaus, S. D. T. .Kravtsov, P. 12 44 .P Mahapatra, P. D. .D Koetke, D. D. 34 .Stringfellow, B. 4 42 .Pandit, Y. 24 .Sandweiss, J. 56 18 55 4 .G Grebenyuk, G. O. .Tang, Z. 40 21 .W Harris, W. J. 39 .Ahammed, Z. 21 44 .Sharma, B. .Surrow, B. .Schmitz, N. 47 .Fisyak, Y. .Smirnov, D. .Lauret, J. 40 9 14 22 .Kesich, A. .S McShane, S. T. 56 .L Ray, L. R. .L Romero, L. J. 17 24 56 .Li, X. 33 49 .Xiao, Z. .Wang, F. 14 57 .Eun, L. .C Bland, C. L. .G Judd, G. E. 48 .Huang, B. 56 55 44 .M .Vasconcelos, S. M. G. 4 .Yang, Y. .Christie, W. .Ullrich, T. .H Zhu, H. Y. 6 .Tokarev, M. .Pluta, J. 19 4 .Aschenauer, E. .J Betancourt, J. M. 47 42 .Bültmann,S. .K Pruthi, K. N. .M Zhang, M. W. .M Du, M. C. 22 3 29 37 8 .D Westfall, D. G. .G Munhoz, G. M. .V Nogach, V. L. .A .Oliveira, N. A. R. 4 .H Tarini, H. L. .Dhamija, S. .Panebratsev, Y. 4 51 4 56 .Krueger, K. 5 .Li, X. 49 34 26 25 .Lu, Y. 6 52 27 56 .Lebedev, A. 37 47 37 4 .A Gagliardi, A. C. .Cui, X. 52 .Kollegger, T. .Sangaline, E. .Kettler, D. 9 .Sharma, M. .Xie, W. .N Svirida, N. D. .Evdokimov, O. 16 .Smirnov, N. 4 .F Chen, F. H. .R Schuster, R. T. .P Hays-Wehle, P. J. 53 51 .Redwine, R. .A .Suaide, P. A. A. .Wang, G. .Yepes, P. .V Alakhverdyants, V. A. 6 4 5 44 .Z Huang, Z. H. 4 25 .Majka, R. 43 12 4 .F Ross, F. J. .G Underwood, G. D. .Plyku, D. .Kabana, S. 24 .G Steadman, G. S. 42 .G Bordyuzhin, G. I. 21 .Zoulkarneeva, Y. .Grosnick, D. .Chung, P. .Li, Y. .Mioduszewski, S. 42 .E Dunkelberger, E. L. 32 .Luo, X. 34 .Trentalange, S. 37 4 22 18 .Dvl Leyva, Davila A. 55 2 .Bunzarov, I. 28 .Przybycien, M. .S Averichev, S. G. 36 4 .Xin, K. 41 .P Zhang, P. X. 54 .Didenko, L. .Kulakov, I. 40 7 26 .Tarnowsky, T. 49 55 6 .WitnJr., Whitten C. .Lednicky, R. .Novak, J. t 56 15 21 42 8 6 .Wang, H. 56 .K Mustafa, K. M. .P Kikola, P. D. 32 26 .Yi, Y. .R Betts, R. R. 41 .Cebra, D. 19 .M Li, M. Z. .S Shi, S. S. .Videbæk, F. 9 12 .Sarkar, A. √ 45 46 10 .Pawlak, T. .Solanki, D. .Konzer, J. .H Chen, H. J. 7 .I Mall, I. O. 14 .Poljak, N. 15 .Luszczak, A. .Reed, R. .Olson, D. .J .Symons, M. J. T. s .Huck, P. .Ruan, L. .Kang, K. .R Gangadharan, R. D. 41 .Fatemi, R. 51 40 NN .Chwastowski, J. 26 .Seele, J. 37 26 .Gupta, S. .C Suarez, C. M. .Xu, H. .Heppelmann, S. 28 21 21 25 19 2 7 28 .Yip, K. . 9GeV 39 - 7.7 = 4 46 ...Stephans, G.S.F. 9 49 6 9 .VnBuren, Van G. .E Tribble, E. R. 21 .Ding, F. .P Burton, P. T. .Odyniec, G. n .Zyzak M. and .Kumar, L. .Borowski, W. 1 6 21 .S Wang, S. J. .M Lima, M. L. 47 17 .Cendejas, R. .Y Shou, Y. Q. 37 .K Mitrovski, K. M. 9 10 25 21 4 6 28 53 .Zhang, Y. 39 57 7 37 .R Pujahari, R. P. 4 49 .K Riley, K. C. .Alekseev, I. 44 26 .J Humanic, J. T. 24 7 .H Lee, H. J. .Rusnak, J. .Margetis, S. .C eSilva, De C. L. .Schambach, J. .C Dunlop, C. J. .Koralt, I. 23 .P Viyogi, P. Y. .Ostrowski, P. .Balewski, J. .Bhasin, A. 37 .Thein, D. .Pawlik, B. .Sorensen, P. 11 .Porter, J. .Kiryluk, J. .Kapitan, J. .Wieman, H. .Y Chen, Y. J. 4 .Seger, J. 20 .Xu, N. .Fazio, S. .Naglis, M. -.Yoo, I-K. .L Ma, L. G. 6 .Guryn, W. .Dion, A. 10 11 25 eso 6.0 version 22 42 25 24 25 41 35 44 4 32 46 31 7 4 45 47 12 19 14 25 25 4 56 33 20 4 22 26 25 26 14 38 4 9 52 25 44 17 25 4 53 47 4 48 31 4 4 2

(STAR Collaboration) 1AGH University of Science and Technology, Cracow, Poland 2Argonne National Laboratory, Argonne, Illinois 60439, USA 3University of Birmingham, Birmingham, United Kingdom 4Brookhaven National Laboratory, Upton, New York 11973, USA 5University of California, Berkeley, California 94720, USA 6University of California, Davis, California 95616, USA 7University of California, Los Angeles, California 90095, USA 8Universidade Estadual de Campinas, Sao Paulo, Brazil 9Central China Normal University (HZNU), Wuhan 430079, China 10University of Illinois at Chicago, Chicago, Illinois 60607, USA 11Cracow University of Technology, Cracow, Poland 12Creighton University, Omaha, Nebraska 68178, USA 13Czech Technical University in Prague, FNSPE, Prague, 115 19, Czech Republic 14Nuclear Physics Institute AS CR, 250 68 Reˇz/Prague,ˇ Czech Republic 15University of Frankfurt, Frankfurt, Germany 16Institute of Physics, Bhubaneswar 751005, India 17Indian Institute of Technology, Mumbai, India 18Indiana University, Bloomington, Indiana 47408, USA 19Alikhanov Institute for Theoretical and Experimental Physics, Moscow, Russia 20University of Jammu, Jammu 180001, India 21Joint Institute for Nuclear Research, Dubna, 141 980, Russia 22Kent State University, Kent, Ohio 44242, USA 23University of Kentucky, Lexington, Kentucky, 40506-0055, USA 24Institute of Modern Physics, Lanzhou, China 25Lawrence Berkeley National Laboratory, Berkeley, California 94720, USA 26Massachusetts Institute of Technology, Cambridge, MA 02139-4307, USA 27Max-Planck-Institut f¨ur Physik, Munich, Germany 28Michigan State University, East Lansing, Michigan 48824, USA 29Moscow Engineering Physics Institute, Moscow Russia 30National Institute of Science and Education and Research, Bhubaneswar 751005, India 31Ohio State University, Columbus, Ohio 43210, USA 32Old Dominion University, Norfolk, VA, 23529, USA 33Institute of Nuclear Physics PAN, Cracow, Poland 34Panjab University, Chandigarh 160014, India 35Pennsylvania State University, University Park, Pennsylvania 16802, USA 36Institute of High Energy Physics, Protvino, Russia 37Purdue University, West Lafayette, Indiana 47907, USA 38Pusan National University, Pusan, Republic of Korea 39University of Rajasthan, Jaipur 302004, India 40Rice University, Houston, Texas 77251, USA 41Universidade de Sao Paulo, Sao Paulo, Brazil 42University of Science & Technology of China, Hefei 230026, China 43Shandong University, Jinan, Shandong 250100, China 44Shanghai Institute of Applied Physics, Shanghai 201800, China 45SUBATECH, Nantes, France 46Texas A&M University, College Station, Texas 77843, USA 47University of Texas, Austin, Texas 78712, USA 48University of Houston, Houston, TX, 77204, USA 49Tsinghua University, Beijing 100084, China 50United States Naval Academy, Annapolis, MD 21402, USA 51Valparaiso University, Valparaiso, Indiana 46383, USA 52Variable Energy Cyclotron Centre, Kolkata 700064, India 53Warsaw University of Technology, Warsaw, Poland 54University of Washington, Seattle, Washington 98195, USA 55Wayne State University, Detroit, Michigan 48201, USA 56Yale University, New Haven, Connecticut 06520, USA 57University of Zagreb, Zagreb, HR-10002, Croatia (Dated: July 17, 2018)

A systematic study is presented for centrality, transverse momentum (pT ) and pseudorapidity (η) dependence of the inclusive charged hadron elliptic ﬂow (v2) at midrapidity ( η 1.0) in Au+Au S S < collisions at √sNN = 7.7, 11.5, 19.6, 27 and 39 GeV. The results obtained with diﬀerent methods, 3

including correlations with the event plane reconstructed in a region separated by a large pseudorapidity gap and 4-particle cumulants (v2 4 ), are presented in order to investigate non-flow { } correlations and v2 fluctuations. We observe that the difference between v2 2 and v2 4 is smaller { } { } at the lower collision energies. Values of v2, scaled by the initial coordinate space eccentricity, v2 ε, ~ as a function of pT are larger in more central collisions, suggesting stronger collective flow develops in more central collisions, similar to the results at higher collision energies. These results are compared to measurements at higher energies at the Relativistic Heavy Ion Collider (√sNN = 62.4 and 200 GeV) and at the Large Hadron Collider (Pb + Pb collisions at √sNN = 2.76 TeV). The v2 pT ( ) values for fixed pT rise with increasing collision energy within the pT range studied ( 2 GeV c). A comparison to viscous hydrodynamic simulations is made to potentially help understand< the energy~ dependence of v2 pT . We also compare the v2 results to UrQMD and AMPT transport model calculations, and( physics) implications on the dominance of partonic versus hadronic phases in the system created at Beam Energy Scan (BES) energies are discussed.

PACS numbers: 25.75.Ld, 25.75.Dw

I. INTRODUCTION azimuthal anisotropies v2 pT of charged hadrons in Au + Au collisions in the sNN( range) of 62.4 - 200 GeV is a √ Azimuthal anisotropies of particle distributions rela- signature of a mixed phase [15]. The new data presented tive to the reaction plane (plane subtended by the impact in this paper shows to what extent such a saturation ef- parameter and beam direction) in high energy heavy- fect is observed. ion collisions have been used to characterize the colli- Several analysis methods for v2 have been pro- sion dynamics [1–3]. In a picture of hydrodynamic ex- posed [30–34]. These are found to be sensitive in varying pansion of the system formed in the collisions, these degrees to non-flow contributions (e.g. correlations due anisotropies are expected to arise due to initial pres- to jets, resonances, etc.) and flow fluctuations. v2 measure gradients and subsequent interactions of the con- surements from various methods have been judiciously stituents [4, 5]. Specifically, differential measurements [6– used to constrain these contributions, in addition to pro- 19] of azimuthal anisotropy have been found to be sensi- viding estimates of systematic errors associated with the tive to (a) the equation of state (EOS), (b) thermaliza- measurements [35]. This is particularly useful for inter- tion, (c) transport coefficients of the medium, and (d) ini- preting results of identified hadron v2 values where, due tial conditions in the heavy-ion collisions. Hence it is im- to limitations of event statistics, it is not possible to use portant to study the dependence of azimuthal anisotropy all methods for v2 analysis. The measurements over a as a function of several variables, for example center-of- range of energies may provide insights to the evolution mass energy ( sNN ), collision centrality, transverse mo- √ of non-flow and flow fluctuations as a function of collision mentum (pT ), and pseudorapidity (η). energy. Recently a beam-energy scan (BES) program has be- gun at RHIC to study the QCD phase diagram [20]. The Inclusive charged hadron elliptic flow measurements at BES program extends the baryonic chemical potential top RHIC energies have been one of the most widely studied observables from the theoretical perspective. It (µB) reach of RHIC from 20 to about 400 MeV [21, 22]. The baryon chemical potential decreases with the de- has been shown that transport models, which provide crease in the beam energy while the chemical freeze-out a microscopic description of the early and late non- temperature increases with increase in beam energy [23]. equilibrium stages of the system, significantly underpre- v This allows one to study azimuthal anisotropy at midra- dict 2 at top RHIC energies, while the inclusion of pidity with varying net-baryon densities. Lattice QCD partonic effects provides a more satisfactory explana- calculations suggest that the quark-hadron transition is tion [36]. The new data discussed here will provide an a crossover for high temperature (T ) systems with small opportunity to study the contribution of partonic matter s and hadronic matter to the v2 measurements as a func- µB or high NN [24]. Several model calculations sug- s √ s tion of NN or (T , µB) by comparisons with models. gest that at larger values of µB or lower NN the transi- √ tion is expected to be first order [25–27].√ Theoretical cal- In this paper we present measurements of the second culations suggest a non-monotonic behavior of v2 could harmonic azimuthal anisotropy using data taken in the be observed around this “softest point of the EOS” [28]. BES program from sNN = 7.7 to 39 GeV. We discuss The softest point of the EOS is usually referred to as the detectors used in√ the analysis, data selections and the temparature/time during which the velocity of sound methods used to determine inclusive charged hadron v2 has a minimum value (or reduction in the pressure of in Sections II and III. Section IV gives v2 results for in- the system) during the evolution. Non-monotonic vari- clusive charged hadrons from different analysis methods. ation of azimuthal anisotropy as a function of collision We discuss the centrality, η, p and sNN dependence of T √ centrality and sNN could indicate the softest point of v2 in Section V, and compare to calculations from trans- the EOS in heavy-ion√ reactions [29]. Further it has been port models. Finally, a summary of the analysis is pre- argued that the observation of saturation of differential sented in Section VI. 4

II. EXPERIMENTS AND DATA SETS to be less than 2 cm. We require that the TPC and FT- PCs have a number of fit points used for reconstruction A. STAR detector of the tracks to be 15 and 5, respectively. For the TPC and FTPCs the> ratio of the> number of fit points to The results presented here are based on data collected maximum possible hits is 0.52. An additional trans- during the tenth and eleventh RHIC runs (2010 and verse momentum cut (0.2 >pT 2 GeV c) is applied to 2011) with the STAR detector using minimum-bias trig- the charged tracks for the< TPC< and FTPC~ event plane gers (requiring a combination of signals from the Beam- determination. Beam Counters (BBC) [37], Zero Degree Calorimeters (ZDC) [38], and Vertex Position Detectors (VPD) [39]). For the 7.7 and 11.5 GeV data, at least one hit in the full barrel Time-Of-Flight detector [40] was required in order to further reduce the background. The main Time C. Centrality determination Projection Chamber (TPC) [41] and two Forward Time Projection Chambers (FTPCs) [42] were used for particle tracking in the central region ( η 1.0) and for- The centrality classes are defined based on the uncor- N raw ward regions (2.5 η 4.0) respectively.S S < Both the rected charged particle multiplicity ( ch ) distribution in the TPC for pseudorapidity η 0.5 and full azimuth. TPC and FTPCs< provided S S < azimuthal acceptance over 2π. The BBC detector subsystem consists of two detec- raw S S < Figure 1 shows the Nch distribution for charged par- tors mounted around the beam pipe, each located outside ticles from the data at sNN = 7.7, 11.5, 19.6, 27 and 39 the STAR magnet pole-tip at opposite ends of the TPC GeV compared to those√ from Monte Carlo (MC) Glauber approximately 375 cm from the center of the nominal simulations. The detailed procedures to obtain the simu- interaction point. Each BBC detector consists of hexag- lated multiplicity are similar to that described in [43]. A onal scintillator tiles arranged in four concentric rings two-component model [44] is used to calculate the simu- that provided full azimuthal coverage. The inner tiles of lated multiplicity distribution given by the BBCs, with a pseudorapidity range of 3.8 η 5.2 were used to reconstruct the event plane in one< S ellipticS < flow analysis. dNch Npart npp 1 x xNcoll , (1) dη 2 Wη=0 = ( − ) + B. Event and track selection

where Npart is the number of participant nucleons and Events for analysis are selected based on collision ver- Ncoll is the number of binary nucleon-nucleon collisions tex positions within 2 cm of the beam axis to reduce in the simulations. The fitting parameter npp is the aver- contributions from beam-gas and beam-pipe (at a radius age multiplicity per unit of pseudorapidity in minimum- of 4 cm) interactions, and within a limited distance from bias p + p collisions and x is the fraction of production the center of the detector along the beam direction ( 70 from the hard component. The inelastic nucleon-nucleon inel cm for the 7.7 GeV data set, 50 cm for the 11.5 GeV± cross section σNN is extracted from fitting the results data set, and 40 cm for the 19.6,± 27 and 39 GeV data of available data for total and elastic p + p cross sec- sets). These values± are chosen to reduce systematics due tions from the Particle Data Group [45]. The x value is to variance in detector performance over η 1.0 while fixed at 0.12 0.02 based on the linear interpolation of s retaining sufficient statistics. After qualityS S cuts, < about the PHOBOS± results at NN = 19.6 and 200 GeV [46]. √ 4 million 0 80% central events remain for 7.7 GeV, 11 Systematic errors on npp are evaluated by varying both million for 11.5− GeV, 20 million for 19.6 GeV, 40 million npp and x within the quoted x uncertainty to determine 2 for 27 GeV and 120 million for 39 GeV data sets. The the minimum χ to describe the data. Since the npp and results from more peripheral collisions are not presented x are anti-correlated, lower (higher) npp is used for higher due to trigger inefficiencies at low multiplicity. The cen- (lower) x for systematic error evaluations on Npart. Ta- trality was defined using the number of charged tracks ble I summarizes the parameters in the two-component inel with quality cuts similar to those in Ref. [12]. The de- model and σNN in the MC Glauber simulations. The tails of the centrality determination will be discussed in event-by-event multiplicity fluctuations are included us- subsection C. The 0 80% central events for v2 analysis ing negative binomial distributions [43]. The centrality of charged hadrons are− divided into nine centrality bins: classes are defined by the fractions of geometrical cross 0 5%, 5 10%, 10 20%, 20 30%, 30 40%, 40 50%, section from the simulated multiplicity distributions. For 50− 60%,− 60 70%− and 70 80%.− − − each centrality bin, average quantities are calculated in A− variety of− track quality− cuts are used to select good the MC Glauber simulations for Npart , Ncoll , reaction charged particle tracks reconstructed using information plane eccentricity εRP , participant⟨ eccentricity⟩ ⟨ ⟩ εpart , from the TPC or FTPCs. The distance of closest ap- root-mean-square⟨ participant⟩ eccentricity εpart 2⟨ , and⟩ proach (DCA) of the track to the primary vertex is taken transverse area Spart . Eccentricity and transverse{ } area ⟨ ⟩ 5

(a) 7.7 GeV (b) 11.5 GeV (c) 19.6 GeV 1010-3-3 1010-4-4 -5-5 ) 1010 -6-6 raw ch 1010 -7-7

/dN 1010 -8-8

evts 1010 0(d) 27 GeV 500 (e) 39 GeV 1000 1500 1010-3-3 )(dN Data Glauber MC 1010-4-4 evts 40-80% -5-5 1010 10-40%

(1/N -6-6 1010 0-10% 1010-7-7 1010-8-8 0 200 400 5000 200 40010000 200 1500 400 raw Nch

raw FIG. 1: (Color online) Distribution of uncorrected multiplicity N measured within η 0.5 in the TPC from √sNN = 7.7 ch S S < to 39 GeV in Au + Au collisions shown as black points. The red curves show the multiplicity distributions at √sNN = 7.7 to 39 GeV from MC Glauber simulations. See texts for more details about simulations. are defined by energies. 2 2 σy σx εRP , (2) σ2 − σ2 III. ELLIPTIC FLOW METHODS = x y +σ2 σ2 2 σ2 y x 4 xy 2 εpart » ,εpart 2 ε , (3) A. The event plane method ( −σ2 )σ2+ ¼ part = x y { } = ⟨ ⟩ 2 2 + 2 Spart π¼σxσy σxy, (4) The event plane method [30] correlates each particle with the event plane determined from the full event mi- σ2 = x2 x−2, σ2 y2 y 2, x y (5) nus the particle of interest, which can be done for each σxy = {xy}−{x} y , ={ }−{ } (6) harmonic. For any Fourier harmonic, n, the event flow where the= curly { }−{ brackets}{ denote} the average over all par- vector (Qn) and the event plane angle (Ψn) are defined ticipants per event, and x and y are the positions of by [30] participant nucleons. Systematic uncertainties on those Qn cos nΨn Qnx wi cos nφi, (7) quantities are evaluated by varying parameters for the i two-component model and by varying the input parame- = = Q ters in the MC Glauber model. The quoted errors are the quadratic sum of the individual systematic uncertainties. Qn sin nΨn Qny wi sin nφi, (8) i Table II summarizes the centrality classes as well as the = = Q

−1 Qny inel Ψn tan n, (9) TABLE I: Summary of npp and σ with systematic uncer- NN Qnx tainties at √sNN = 7.7, 11.5, 19.6, 27 and 39 GeV. x is set to = ~ 0.12 0.02 for all collision energies. where sums extend over all particles i used in the event ± inel plane calculation, and φi and wi are the laboratory az- √sNN (GeV) npp σNN (mb) imuthal angle and the weight for the ith particle, respec- 7.7 0.89 0.04 30.8 1.0 ± ± tively. The reaction plane azimuthal distribution should 11.5 1.07 0.05 31.2 1.0 ± ± be isotropic or flat in the laboratory frame if the detectors 19.6 1.29 0.05 32.0 1.0 ± ± have ideal acceptance. Since the detectors usually have 27 1.39 0.06 33.0 1.0 ± ± non-uniform acceptance, a procedure for flattening the 39 1.52 0.08 34.0 1.0 laboratory event plane distribution is necessary [47, 48]. ± ± As shown in Eq. (10), the observed v2 is calculated results obtained by MC Glauber simulations at the five with respect to the reconstructed event plane angle Ψn 6 where n equals 2 when we use the second harmonic event large acceptance losses from hardware faults caused sig- plane and n equals 1 when we use the first harmonic event nificant gaps in the azimuthal angle distribution of these plane. tracks, preventing use of the φ weight method because obs of the inability to define φ weights in regions of zero ac- v2 cos 2 φ Ψn (10) [ ( )]⟩ ceptance. Thus, only the re-centering method is used for The angular brackets= indicate ⟨ an− average over all parti- the FTPC. cles in all events. However, tracks used for the v2 calculation are excluded from the calculation of the flow vector to remove self-correlation effects. Because the estimated 3. BBC event plane reaction plane fluctuates due to finite number of particles, one has to correct for this smearing by dividing the In this method the first-order event plane is recon- observed correlation by the event-plane resolution (the structed using particle trajectories determined from hits denominator in Eq. (11)), which is the correlation of the in the BBC detectors. In this case, φi denotes the fixed event plane with the reaction plane. azimuthal angle of the center of the ith BBC tile in Eq. (7) obs and (8), and wi is the fraction of BBC-observed energy v2 v2 (11) deposition recorded in tile i: cos 2 Ψn Ψr = ⟨ [ ( )]⟩ Since the reaction plane is unknown,− the denominator Ai wi . (13) in Eq. (11) could not be calculated directly. As shown in Ai = Eq. (12), we estimate the event plane resolution by the ∑ correlation between the azimuthal angles of two subset The BBC event plane obtained from one BBC detector groups of tracks, called sub-events A and B. In Eq. (12) is called a sub-event. A combination of the sub-event C is a factor calculated from the known multiplicity de- plane vectors for both BBC detectors provides the full pendence of the resolution [30]. event plane. cos 2 Ψ Ψ C cos 2 ΨA ΨB (12) cos 2 φ Ψ1 n r » n n v2 BBC (14) ⟨ [ ( )]⟩ ⟨ [ ( )]⟩ ⟨ [ ( −A )]⟩B − = − { } C» cos 2 Ψ1 Ψ1 Random sub-events are used for TPC event plane, while = ⟨ [ ( )]⟩ pseudorapidity sub-events are used for FTPC/BBC event − where C is the constant in Eq. (12). ΨA ,ΨB are sub- plane. 1 1 event plane angles from each BBC detector and Ψ1 is the full event plane angle from both sub-events combined. 1. TPC event plane The detector acceptance bias is removed by applying the shift method [48]. Equation (15) shows the formula for the shift correction. The averages in Eq. (15) are The TPC event plane means the event plane recon- taken from a large sample of events. In this analysis, the structed from tracks recorded by the TPC. For this event correction is done up to the twentieth harmonic. The plane the φ weight method is an effective way to flatten distributions of ΨA and ΨB are separately flattened and the azimuthal distribution for removing detector accep- 1 1 then the full-event event plane distribution is flattened. tance bias. These weights are generated by inverting the Accordingly, the observed v and resolution are calcu- φ distributions of detected tracks for a large event sam- 2 lated using the shifted (sub)event plane azimuthal angles. ple. The φ weights are folded into the weight wi in Eq. (7) and Eq. (8). ′ 1 Ψ Ψ n n sin 2nΨ cos 2nΨ The re-centering correction [47, 48] is another method [ ⟨ ( )⟩ ( ) to calibrate the event plane. In this method, one sub- = + ∑ − (15) tracts from the Q-vector of each event the Q-vector av- cos 2nΨ sin 2nΨ eraged over many events. For both the φ weight and + ⟨ ( )⟩ ( )] re-centering methods, the corrections are applied in each More details for the BBC event plane have been described centrality bin, in 2 bins of the primary vertex position in Ref. [49]. along the longitudinal beam direction (Vz), and in 2 bins for positive/negative pseudorapidity. These corrections are determined as a function of data collection time. The B. The η sub-event method difference in the effects on v2 from the different flattening techniques is negligible. The η sub-event method is similar to the event plane method, except one defines the flow vector for each particle based on particles measured in the opposite hemi- 2. FTPC event plane sphere in pseudorapidity:

Forward-going tracks reconstructed in the two FTPCs cos 2 φ± Ψ2,η∓ v2 EtaSubs ⟨ [ ( )]⟩ (16) can also be used to determine the event plane. However, − { } = » cos 2 Ψ2,η+ Ψ2,η− ⟨ [ ( − )]⟩ 7

Here v2 EtaSubs denotes the results of the η sub- 2. The Q-cumulants method event method{ and Ψ} 2,η+ (Ψ2,η− ) is the second harmonic event plane angle determined by particles with positive The Q-cumulants method [50] is a recent method (negative) pseudorapidity. An η gap of η 0.075 is used to calculate cumulants without using nested loops over between negative (positive) η sub-eventS S to< reduce non- tracks and without generating functions [33]. The ad- flow correlations between the two ensembles. vantage is that it provides fast (one loop over data) and exact non-biased (no approximations and no interference between different harmonics) estimates of the correlators C. The cumulant method compared to the generating function cumulants. The cumulants are expressed in terms of the moments of the The advantage of the cumulant method is that the magnitude of the corresponding flow vector Qn multi-particle cumulant is a higher-order multi-particle correlation formalism which removes the contribution of M inφi non-flow correlations from lower-order correlations [32, Qn e (22) 33]. The measured 2-particle correlations can be ex- ≡ iQ=1 pressed with flow and non-flow components: The single-event average two- and four-particle azimuthal ein(φ1−φ2) ein(φ1−Ψr) ein(Ψr −φ2) δ correlations can be then formulated as: n (17) ⟨ ⟩ ⟨ 2 ⟩⟨ ⟩ 2 = v δn + Q − M n 2 n (23) = + MS MS − 1 Here n is the harmonic number and δn denotes the non- ⟨ ⟩ = flow contribution. The average should be taken for all ( ) pairs of particles in a certain rapidity and transverse mo- Q 4 + Q 2 − 2 ⋅ R Q Q∗ Q∗ mentum region, and for all events of a data sample. The 4 n 2n 2n n n S MS MS − 1S M − 2[ M − 3 ] measured 4-particle correlations can be expressed as: ⟨ ⟩ = (24) (M − )(⋅ Q 2 −)(M M −) in(φ1+φ2−φ3−φ4) 4 ⋅ ⋅ 2 + 2 (18) − 2 2 n 3 e vn 2 2 vnδn 2δn 2 ( ) S S ( ) ⟨ ⟩ M M − 1 M − 2 M − 3 Thus the flow contribution= can+ be obtained by subtract- ( )( )( ) ing the 2-particle correlation from the 4-particle correla- The average over all events can be performed as: tion: in(φ1−φ2) ein(φ1+φ2−φ3−φ4) ein(φ1+φ2−φ3−φ4) − 2 e (19) ⟪ ⟫ ⟪ ⟫ ⟨⟨ ⟩⟩ ⟨in(φ1−φ3) 2 − ⟩4 ≡ W =2 e vn event ⟨2⟩ i 2 i (25) ⟨ ⟩ ∑ ⟨ ⟩ where ... is used for the cumulant. The= cumulant of event W⟨2⟩ ≡ i order two⟨⟨ is⟩⟩ just ein(φ1−φ2) ein(φ1−φ2) . ∑ ⟨⟨ ⟩⟩ = ⟨ ⟩ 4 ein(φ1+φ2−φ3−φ4) 1. The cumulant method with generating function ⟪ ⟫ ≡ ⟪ ⟫ events W⟨4⟩ 4 i (26) i ⟨ ⟩ ∑ W The GF-cumulant method is computed from a gener- ≡ events ⟨4⟩ i ∑ ating function [33]: while the weights are the number of two- and four- M particle combinations: wj ∗ inφj −inφj Gn z 1 + z e + ze (20) ( ) j=1[ M ( )] − = M W⟨2⟩ M M 1 , (27) Here z is an arbitrary complex number, z∗ denotes its ≡ ( ) M complex conjugate, denotes the multiplicity in each − − − event, and w is the weight (transverse momentum, ra- W⟨4⟩ M M 1 M 2 M 3 . (28) j ( )( )( ) pidity etc.) used in the analysis. The event-wise averaged ≡ Choosing the multiplicity weights above can make the generating function then can be expanded in powers of final multi-particle azimuthal correlations free of mul- z and z∗ where the coefficients of expansion yield the tiplicity fluctuations [51]. However, one can also use correlations of interest: unit weights treating events with different multiplicity −inφ1 ∗ inφ1 Gn z 1 + z e + z e + equally. The two- and four-particle cumulants without 2 ∗2 ⟨ ( )⟩ M−1 z ⟨ −in(φ⟩1+φ2) ⟨ + z ⟩ in(φ1+φ2) detector bias then can be formulated as: =M 2 e 2 e (21) +zz∗(ein⟨(φ1−φ2) + ⟩... ⟨ ⟩ ⟨ ⟩) cn 2 2 (29) These correlations can be used to construct the cumu- { } = ⟪ ⟫ lants. More details for the analysis of STAR data have − ⋅ 2 been described in Ref. [10]. cn 4 4 2 2 (30) { } = ⟪ ⟫ ⟪ ⟫ 8

1 nd 1 (a) TPC 2 harmonic Event Plane (b) FTPC (BBC) Event Plane 39 GeV 0.8 0.8 27 GeV 19.6 GeV 0.6 0.6 11.5 GeV 7.7 GeV

0.4 0.4 Resolution 0.2 0.2

0 0 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 % Most Central

FIG. 2: (Color online) The event plane resolutions for Au + Au collisions at √sNN = 7.7, 11.5, 19.6, 27 and 39 GeV as a function of collision centrality. Panel (a) shows the resolution of the second harmonic event plane from the TPC ( η 1). Panel (b) shows the resolution for second harmonic event plane from the FTPCs (2.5 η 4.0) for 39 GeV and secondS S < harmonic event plane resolution correction using the ﬁrst-order event plane from the BBCs (3<. S8S <η 5.2) for 7.7, 11.5, 19.6 and 27 GeV. < S S <

The reference flow (e.g. integrated over pT ) can be The event average can be obtained as follows: estimated both from two- and four-particle cumulants: ′ events w⟨2′⟩ i 2 i 2′ (37) v 2 »c 2 (31) N ( ) ⟨ ⟩ n n ⟪ ⟫ ∑ w⟨2′⟩ i { } { } = i=1 = ∑ ( ) 4 − ′ vn 4 » cn 4 (32) w 4′ i 4 i { } { } ′ events ⟨ ⟩ = 4 N ( ) ⟨ ⟩ (38) ∑ ′ Once the reference flow is estimated, we proceed to the ⟪ ⟫ i=1 w⟨4 ⟩ i = ( ) calculation of differential flow (e.g. as a function of pT ) ∑ of the particle of interest (POI) which needs another two Multiplicity weights are: vectors p and q. Particles used to estimate reference flow w ′ m M − m (39) are called reference particles (REP). For particles labeled ⟨2 ⟩ p q as POI: ≡ − − − mp w⟨4′⟩ mpM mq M 1 M 2 (40) inψi pn e . (33) ≡ ( )( )( ) ≡ iQ=1 The two- and four-particle differential cumulants without detector bias are given by: For particles labeled as both POI and REP: ′ mp dn 2 2 (41) inψi { } ⟪ ⟫ qn e . (34) = i=1 ≡ Q ′ ′ dn 4 4 − 2 ⋅ 2 2 (42) Then the reduced single-event average two- and four- { } = ⟪ ⟫ ⟪ ⟫⟪ ⟫ particle correlations are: Equations for the case of detectors without uniform ac- ∗ ceptance can be found in Ref. [50]. Estimations of differ- pnQ − mq 2′ n (35) ential flow are expressed as: m M − m ⟨ ⟩ = p q ′ dn 2 vn 2 { } (43) ′ ∗ ∗ − ∗ ∗ − ∗ { } »cn 2 4 pnQnQnQn q2nQnQn pnQnQ2n = ⟨ ⟩ [ { } = − 2 ⋅ Mp Q∗ − 2 ⋅ m Q 2 + 7 ⋅ q Q∗ n n q n n n (36) − ∗ + ∗ + S⋅ S ∗ + ⋅ d 4 Qnqn q2nQ2n 2 pnQn 2 mqM v′ 4 n (44) n { }3~4 − ⋅ − − − −cn 2 6 mq mpM 3mq M 1 M 2 { } = ]~[( )( )( )] { } 9

(a) 0-5% (b) 5-10% (c) 10-20% 39 GeV v {2}{GF} 0.2 2 0.2 0.2 v2{4}{GF} v2{2}{Q} v2{4}{Q} 0.1 0.1 0.1 2

v 0 0 0 0(d) 20-30% 1 2 3 4 0(e) 30-40% 1 2 3 4 0(f) 40-50% 1 2 3 4 0.2 0.2 0.2

0.1 0.1 0.1

0 0 0 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 p (GeV/c) T

FIG. 3: (Color online) The comparison of v2 as a function of pT between GF-cumulant (open symbols) and Q-cumulant (full symbols) methods in Au+Au collisions at √sNN = 39 GeV. v2 4 fails in most central (0 5%) collisions due to the small { } − values of v2 and large v2 ﬂucuations.

(a1) 7.7 GeV (b1) 11.5 GeV (c1) 19.6 GeV (d1) 27 GeV (e1) 39 GeV 0.08 0.08 0.08 0.08 0.08

0.06 0.06 0.06 0.06 0.06 2 0.04 0.04 0.04 0.04 v 0.04

0.02 0.02 0.02 v2{BBC} 0.02 v2{2} 0.02 v2{EP} v2{FTPC} v2{4} v2{EtaSubs}

0(a2) 20 Ratio 40 to v 60{2} 80 (b2) Ratio to v {2} (c2) Ratio to v {2} (d2) Ratio to v {2} (e2) Ratio to v {2} 1.2 2 1.2 2 1.2 2 1.2 2 1.2 2

1 1 1 1 1

0.8 0.8 0.8 0.8 0.8 Ratio 0.6 0.6 0.6 0.6 0.6 0 20 40 60 80 0 20 40 60 80 0 20 40 60 80 0 20 40 60 80 0 20 40 60 80 % Most Central

FIG. 4: (Color online) The pT ( 0.2 GeV/c) and η ( η 1) integrated v2 as a function of collision centrality for Au + Au > S S < collisions at √sNN = 7.7 GeV (a1), 11.5 GeV (b1), 19.6 GeV (c1) , 27 GeV (d1) and 39 GeV (e1). The results in the top panels are presented for several methods of obtaining v2. The bottom panels show the ratio of v2 obtained using the various techniques, with respect to v2 2 . The error bars shown are statistical. { }

IV. RESULTS different detectors and the cumulant method are used in the analysis. The event planes are determined from A. The event plane resolution the TPC in the midrapidity region, and the FTPC/BBC at forward rapidity. The η gap between FTPC/BBC to TPC could reduce the non-flow contribution in the v2 To investigate the non-flow correlations and v2 fluctuations of the v2 measurements, the event planes from 10

(a1) 7.7 GeV (b1) 11.5 GeV (c1) 19.6 GeV (d1) 27 GeV (e1) 39 GeV 0.2 0.2 0.2 0.2 0.2 0.15 0.15 0.15 0.15 0.15

2 0.1 0.1 0.1 0.1 0.1 v 0.05 0.05 0.05 0.05 0.05 v2{BBC} v2{2} v2{EP} v {FTPC} v {4} v {EtaSubs} 0 0 0 2 0 2 0 2 0(a2) 1 Ratio 2 to v {2} 3 (b2) Ratio to v {2} (c2) Ratio to v {2} (d2) Ratio to v {2} (e2) Ratio to v {2} 1.2 2 1.2 2 1.2 2 1.2 2 1.2 2

1 1 1 1 1

0.8 0.8 0.8 0.8 0.8 Ratio

0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 p (GeV/c) T

FIG. 5: (Color online) The v2 as a function of pT for 20 30% central Au + Au collisions at midrapidity for √sNN = 7.7 GeV − (a1), 11.5 GeV (b1), 19.6 GeV (c1) , 27 GeV (d1) and 39 GeV (e1). The top panels show v2 vs. pT using various methods as labeled in the figure and discussed in the text. The bottom panels show the ratio of v2 measured using the various methods with respect to v2 2 . { } measurement [13]. Figure 2 shows the event plane res- approximation. The cumulant method (GF-cumulant or olution from TPC (panel (a)) and BBC (FTPC) (panel Q-cumulant) used in the analysis does not cause differ- (b)). The resolution of the TPC second harmonic event ence in the comparison with other experimental results plane increases as the collision energy increases, as the and theoretical calculations. To be consistent with the resolution depends on the multiplicity and the v2 sig- previous STAR results, we will hereafter show only re- nal [30]. Due to limited statistics, the FTPC event plane sults from the GF-cumulant method. is used only for the 39 GeV data set where the BBC event plane cannot be used because of the poor resolu- Other method comparisons are shown in Figs. 4 and tion. The resolution of the FTPC event plane is about 5 for inclusive charged hadrons in Au + Au collisions at s four times lowers than the TPC event plane. The BBC √ NN = 7.7 GeV (a1), 11.5 GeV (b1), 19.6 GeV (c1), is used to determine the event plane for the 7.7, 11.5, 27 GeV (d1) and 39 GeV (e1). As the v2 measurements 19.6 and 27 GeV data sets. Note the BBC event plane is from various methods are obtained using charged tracks based on the first harmonic, as the v1 signal is significant recorded at midrapidity ( η 1), the statistical errors in the rapidity region covered by the BBC. The quali- on the results from the differentS S < v2 methods are thus tively different centrality dependence of the FTPC and correlated. The conclusions on the differences in v2 val- BBC event plane resolutions is because of the different ues from different methods are based on the systematic centrality dependence of v1 and v2. trends observed for the corresponding ratios with respect to v2 2 . Figure 4 shows v2 integrated over 0.2 pT 2.0 GeV/{c }and η 1 versus centrality. For comparison< < pur- poses, the integratedS S < v2 values for all methods are divided B. Method comparison by the values of the 2-particle cumulant method (v2 2 ) and plotted in panels (a2) through (e2). The results{ } of The comparison of v2 as a function of pT between the 4-particle cumulants are systematically lower than the GF-cumulant and Q-cumulant methods is shown in the other methods, except for v2 FTPC BBC . The Fig. 3 for six collision centralities in Au+Au collisions difference is about 10 − 20% in 39,{ 27 and~ 19.6} GeV, − − at √sNN = 39 GeV. The GF-cumulant and Q-cumulant 10 15% in 11.5 GeV and 5 10% in 7.7 GeV. The η methods agree within 5% at all five collision energies. sub-event values for peripheral collisions (50 − 60% to Compared to GF-cumulant method, the recently devel- 70−80%) drop below the 2-particle and TPC event plane oped Q-cumulant is the exact cumulant method [50]. results, indicating the η sub-event method could reduce The observation of consistency between the two meth- some non-flow correlations for peripheral collisions. Non- ods at BES energies implies the GF-cumulant is a good flow correlations are defined as correlations not related to 11

0.8 (a) 7.7 GeV 0.8 (b) 11.5 GeV 0.8 (c) 19.6 GeV 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 {2} 0.2 0 0 0 part (d) 27 GeV (e) 39 GeV ε 0.8 0 1 2 0.8 0 1 2 0.8 0 1 2

⁄ Au + Au collisions 0.6 0.6 0.6 2 50-60% v 0.4 0.4 0.4 30-40% 10-20% 0.2 0.2 0.2 Glauber eccentricity 0 0 0 0 1 2 0 1 2 0 1 2 p (GeV/c) T

FIG. 6: (Color online) The v2 over ε (Glauber) as a function of pT for various collision centralities (10 20%, 30 40% and − − 50 60%) in Au + Au collisions at midrapidity. Panel (a), (b), (c), (d) and (e) show the results for √sNN = 7.7, 11.5, 19.6, − 27 and 39 GeV respectively. The data are from v2 EtaSubs . The error bars and shaded boxes represent the statistical and systematic uncertainties respectively, as described in{ Sec. IV} C.

the reaction plane. The dominant non-flow correlations atively small and less dependent on pT . It suggests the originating from two-particle correlations (such as HBT non-flow contribution to the event plane and 2-particle correlations, resonance decay) scale as 1 N [30], where correlation methods depends on pT . Based on the inter- 2 N is the multiplicity of particles used to~ determine the pretation in Ref. [1], the difference between v2 2 and 2 event plane. Thus the non-flow contribution is larger v2 4 is approximately equal to non-flow plus two{ } times in peripheral collisions. In mid-central and peripheral v2{fluctuations.} The fact that the ratio of v2 4 to v2 2 collisions (10 − 20% to 40 − 50%), the data of v2 BBC is closer to 1 at the lower collision energies indicates{ } { the} from 7.7, 11.5, 19.6 and 27 GeV are consistent with{ v2 4} non-flow and/or v2 fluctuations in the v2 measurement and lower than other methods. It suggests the first-order{ } depend on the collision energy. One possible explanation (BBC) event plane suppresses the second-order non-flow is that the non-flow correlations from jets presumably and/or fluctuation effects. Within statistical errors, the decrease as the collision energy decreases. The results of s results of v2 FTPC from Au + Au collisions at √ NN v2 BBC are found to be consistent with v2 4 in 7.7, = 39 GeV are{ close} to v2 2 , v2 EP and v2 EtaSubs 11.5,{ 19.6} and 27 GeV, while the v2 FTPC{ }is larger in semi-central collisions (10{ }− 20%{ to} 20 − 30%).{ In the} than v2 4 in 39 GeV. This consistency{ can be} also ob- peripheral collisions (30 − 40% to 60 − 70%), v2 FTPC served in{ Fig.} 4 for 10 − 20% to 40 − 50% centrality bins. falls between v2 EtaSubs and v2 4 . It indicates{ that} It indicates that the use of the first-order reaction plane the η gap between{ TPC and} FTPC{ reduces} the non-flow (BBC event plane) to study the second harmonic flow contribution. eliminates flow fluctuations which are not correlated between different harmonics. The first-order BBC reaction The pT differential v2 from various methods for the plane is struck by nucleon spectators for these beam en- − 20 30% centrality bin are shown in the upper panels ergies. The contribution of spectators makes the BBC of Fig. 5. For comparison, the v2 from other meth- event plane more sensitive to the reaction plane. This ods are divided by the results of the 2-particle cumu- could partly explain the consistency between v2 BBC lant method and shown in the lower panels of Fig. 5. and v2 4 mentioned above. More studies of the{ colli-} It can be seen that the difference of v2 2 compared to sion energy{ } dependence of non-flow and flow fluctuations { } v2 FTPC BBC , v2 2 and v2 EtaSubs depends on the will be discussed in another paper. pT{ range.~ A larger} { difference} { can be} observed in the low pT region (pT 1 GeV/c). Beyond pT = 1 GeV/c the difference stays< constant in the measured pT range. The difference between v2 FTPC BBC and v2 4 is rel- { ~ } { } 12

0.8 (a) 7.7 GeV 0.8 (b) 11.5 GeV 0.8 (c) 19.6 GeV 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 {2} 0.2 0 0 0 part (d) 27 GeV (e) 39 GeV ε 0.8 0 1 2 0.8 0 1 2 0.8 0 1 2

⁄ Au + Au collisions 0.6 0.6 0.6 2 50-60% v 0.4 0.4 0.4 30-40% 10-20% 0.2 0.2 0.2 CGC eccentricity 0 0 0 0 1 2 0 1 2 0 1 2 p (GeV/c) T

FIG. 7: (Color online) The v2 over ε (CGC) as a function of pT for various collision centralities (10 20%, 30 40% and − − 50 60%) in Au + Au collisions at midrapidity. Panel (a), (b), (c), (d) and (e) show the results for √sNN = 7.7, 11.5, 19.6, − 27 and 39 GeV respectively. The data are from v2 EtaSubs . The error bars and shaded boxes represent the statistical and systematic uncertainties respectively, as described in{ Sec. IV} C.

C. Systematic uncertainties the distance of closest approach to the primary vertex for the tracks, and the number of fit points used for reconstruction of the tracks). The systematic uncertainties Different v methods show different sensitivities to 2 on v 2 and v 4 are based on the difference between non-flow correlations and v fluctuations. In previ- 2 2 2 Q-cumulant{ } and{ GF-cumulant} methods (5%) as well as ous STAR publications, the differences between differ- cut variations (5%). All the percentage uncertainties are ent methods were regarded as systematic uncertain- relative to the v value. ties [11, 12]. A great deal of progress has revealed that 2 some of these differences are not due to systematic uncertainties in different methods, but due to different sensitivities to non-flow and flow fluctuation effects [35, 52]. The V. DISCUSSION four particle cumulant method is less sensitive to non- flow correlations [32, 33] and has a negative contribution A. Transverse momentum and centrality from flow fluctuations. v2 measurements from the two dependence of v2 particle cumulant method and the event plane method

(the second harmonic event plane) have positive contri- The centrality dependence of pT differential v2 with butions from flow fluctuations as well as non-flow. It was respect to the initial eccentricity has been studied in de- also noticed that four particle cumulant results should s tail for Au + Au and Cu + Cu collisions in √ NN = 200 be very close to flow in the reaction plane, while the and 62.4 GeV [12, 13]. The larger magnitude of v2 in the two particle cumulant measures flow in the participant more peripheral collisions could be due to the larger ini- plane [35, 52]. Further, because of the large pseudora- tial eccentricity in coordinate space for the more periph- pidity gap between the BBC/FTPC and TPC, v2 BBC eral collisions. The participant eccentricity is the initial { } and v2 FTPC are most insensitive to non-flow correla- configuration space eccentricity of the participant nucle- tions. { } ons defined by Eq. (3). The root-mean-square partici- We estimate the systematic uncertainty on event plane pant eccentricity, εpart 2 , is calculated from the Monte flattening methods for v2 EP and v2 EtaSubs by the Carlo Glauber model [53,{ } 54] (Tab. II) and Color Glass difference between them and{ find} it to{ be negligible} (be- Condensate (CGC) model [55–58] (Tab. III). The event low 1%). A 5% systematic uncertainty on v2 BBC , plane is constructed from hadrons which have their ori- v2 FTPC , v2 EP and v2 EtaSubs is estimated{ by} gin in participant nucleons. At the same time, the event varying{ cut} parameters{ } (e.g.{ collision} vertex position, plane resolution (η sub-event) is less than 0.5. Thus, 13

(a) (b) 0.06 0.06 0.05 0.05 2 v 0.04 0.04

0.03 200 GeV 39 GeV0.03 19.6 GeV 7.7 GeV 62.4 GeV 27 GeV 11.5 GeV 1.6 (c)-1 Ratio -0.5 to 0 fit function 0.5 11.6 -0.4(d) Ratio -0.2 to 0fit function 0.2 0.4

1.4 1.4

1.2 1.2 Ratio 1 1

-1 -0.5 0 0.5 1 -0.4 -0.2 0 0.2 0.4 η η ⁄ y beam

FIG. 8: (Color online) Panel (a) shows the v2 EP vs. η for 10 40% centrality in Au + Au collisions at √sNN = 7.7, 11.5, { } − 19.6, 27, 39, 62.4 and 200 GeV. Panel (c) shows the ratio of v2 vs. η for all √sNN with respect to the fit curve. Panel (b) shows the v2 EP vs. η/ybeam. Panel (d) shows the ratio of v2 vs. η/ybeam for all √sNN with respect to the fit curve. The data for { } √sNN = 62.4 and 200 GeV are from refs. [12, 61, 62]. The dashed red curves show the empirical fit to the results from Au + Au collisions at √sNN = 7.7 GeV. The bands show the systematic uncertainties as described in Sec. IV C.

what we actually measure is the root-mean-square of v2 the rest of the collision energies. The 7.7 GeV data are with respect to the participant plane [52]. In this case, empirically fit by the following function: εpart 2 is the appropriate measure of the initial geomet- 2 4 ric anisotropy{ } taking the event-by-event fluctuations into v2 η p0 + p1η + p2η (45) account [52, 59, 60]. In Figs. 6 and 7, the centrality de- ( ) = p . ± . p − . ± pendence of pT differential v2 over eccentricity is shown with parameters 0 0 0450 0 0002, 1 0 0064 s 0.0015, p2 −0.0024 ± 0.0017. For clarity, the panel (c) for Au + Au collisions at √ NN = 7.7, 11.5, 19.6, 27 = = and 39 GeV. For all five collision energies, the centrality of Fig. 8 shows= the ratio of v2(η) with respect to this fit dependence of v (p ) is observed to be similar to that at function. The pseudorapidity dependence of v2 indicates 2 T s higher collision energies (62.4 and 200 GeV) of Au + Au a change in shape as we move from √ NN = 200 GeV to − and Cu + Cu colliding systems. That central collisions 7.7 GeV within our measured range 1 η 1. in general have higher v2 ε than peripheral collisions is To investigate the collision energy dependence< < of the consistent with the picture~ that collective interactions are v2(η) shape, in panel (b) and (d) of Fig. 8, the same stronger in collisions with larger numbers of participants. v2 results have been plotted as a function of pseudorapidity divided by beam rapidity. The data of 7.7 GeV are fit by Eq. (45) with parameters p0 0.0450 ± 0.0002, − ± − ± p1 0.0279 0.0064, p2 0.0464 0=.0325. The beam B. Pseudorapidity dependence rapidities= are 2.09, 2.50, 3.04,= 3.36, 3.73, 4.20 and 5.36 s for √ NN = 7.7, 11.5, 19.6, 27, 39, 62.4 and 200 GeV The panel (a) of Fig. 8 shows v2 as a function of pseu- respectively. After dividing pseudorapidity by the beam s dorapidity for Au + Au collisions at √ NN = 7.7, 11.5, rapidity, the shape of v2 seems similar at all collision en- 19.6, 27, 39, 62.4 and 200 GeV in mid-central (10 − 40%) ergies. The approximate beam rapidity scaling on the s collisions. The data for √ NN = 62.4 and 200 GeV are v2 η shape suggests the change in shape may be related from refs. [12, 61, 62]. To facilitate comparison with 62.4 to( the) final particle density. Higher particle density indi- and 200 GeV data, the results of v2 EP are selected for cates higher probability of interaction which can generate { } 14

(a1) 10 - 20% (b1) 20 - 30% (c1) 30 -40% 0.2 Fit to 200 GeV data0.2 0.2 0.15 0.15 0.15 2 0.1 0.1 0.1 v {4} 2.76 TeV 27 GeV 200 GeV 19.6 GeV 0.05 0.05 62.4 GeV 0.05 11.5 GeV 39 GeV 7.7 GeV 0 0 0 1.4 0(a2) 1Ratio to 2 fit function 3 1.4 (b2) Ratio to fit function1.4 (c2) Ratio to fit function

1.2 1.2 1.2 1 1 1

0.8 0.8 0.8 Ratio 0.6 0.6 0.6 0 1 2 3 0 1 2 3 0 1 2 3 p (GeV/c) T

FIG. 9: (Color online) The top panels show v2 4 vs. pT at midrapidity for various collision energies (√sNN = 7.7 GeV to 2.76 { } TeV). The results for √sNN = 7.7 to 200 GeV are for Au + Au collisions and those for 2.76 TeV are for Pb + Pb collisions. The dashed red curves show the empirical ﬁts to the results from Au + Au collisions at √sNN = 200 GeV. The bottom panels show the ratio of v2 4 vs. pT for all √sNN with respect to the ﬁt curve. The results are shown for three collision centrality classes: 10 20% (a1),{ } 20 30% (b1) and 30 40% (c1). Error bars are shown only for the statistical uncertainties. − − −

(a) MC-Glb, η/s=0.08 (b) MC-KLN, η/s=0.20 1.4 1.4 2.76 TeV 2.76 TeV 27 GeV 20-30% 39 GeV 62.4 GeV 19.6 GeV 11.5 GeV 39 GeV 11.5 GeV 1.2 7.7 GeV 1.2 7.7 GeV

1 1 (200 GeV) 2

v 0.8 0.8 ⁄

2 v 0.6 0.6

0 0.5 1 1.5 2 0 0.5 1 1.5 2 p (GeV/c) T

FIG. 10: (Color online) The experimental data (symbols) are the same as in Fig. 9 (b2). The lines represent the viscous hydrodynamic calculations from Ref. [5] based on (a) MC-Glauber initial conditions and η s = 0.08 (b) MC-KLN initial conditions and η s = 0.20. ~ ~ larger collective ﬂow. C. Energy dependence

One of the most important experimental observations at RHIC is the signiﬁcant v2 signal in the top energy of Au + Au collisions [6, 10] (more than 50% larger than at 15

(a1) 10 - 20% (b1) 20 - 30% (c1) 30 -40% 0.2 0.2 0.2 0.15 0.15 0.15 0.1 0.1 0.1 39 GeV 11.5 GeV 27 GeV 7.7 GeV {EtaSubs} 0.05 0.05 0.05

2 19.6 GeV

v 0 0 0 0(a2) 1Ratio to 2 39 GeV 3 (b2) Ratio to 39 GeV (c2) Ratio to 39 GeV 1.1 1.1 1.1 1 1 1 0.9 0.9 0.9 0.8 0.8 0.8 Ratio 0.7 0.7 0.7 0 1 2 3 0 1 2 3 0 1 2 3 p (GeV/c) T

FIG. 11: (Color online) The top panels show v2 EtaSubs vs. pT at midrapidity for various collision energies (√sNN = 7.7 { } GeV to 39 GeV). The bottom panels show the ratio of v2 EtaSubs vs. pT for all √sNN with respect to the 39 GeV data. The results are shown for three collision centrality classes: 10{ 20% (a1),} 20 30% (b1) and 30 40% (c1). Error bars are shown only for the statistical uncertainties. − − −

the SPS [63]). It could be interpreted as the observation v2 pion v2 for pT below 2 GeV/c, could partly explain of a higher degree of thermalization than at lower colli- the< collision energy dependence. Further, in Fig. 10 we sion energies [6]. The BES data from the RHIC-STAR compare the experimental data from Fig. 9 (b2) to the experiment offers an opportunity to study the collision viscous hydrodynamic calculations [5]. As the collision s energy dependence of v2 using a wide acceptance detec- energy varies from √ NN = 7.7 to 2760 GeV, the exper- tor at midrapidity. Figure 9 shows the pT dependence of imental data show larger splitting in the lower pT region s − v2 4 from √ NN = 7.7 GeV to 2.76 TeV in 10 20% (a1), and converge at the intermediate range (pT 2 GeV/c); 20{−}30% (b1) and 30 − 40% (c1) centrality bins, where while, in the pure viscous hydrodynamic simulations,∼ the the ALICE results in Pb + Pb collisions at √sNN = splitting increases with pT . The pT dependence of the 2.76 TeV are taken from Ref. [18]. The reasons to select v2 ratio cannot be reproduced by pure viscous hydro- the results of v2 4 for the comparison are the following: dynamic simulations with a constant shear viscosity to 1) keep the method{ } for v2 measurements consistent with entropy density ratio (η s), and zero net baryon density. the published results of ALICE; 2) v2 4 is insensitive to The comparison suggests~ that a quantitative study at non-flow correlations. The 200 GeV{ data} is empirically lower collision energies requires a more serious theoret- fit by a fifth order polynomial function. The parameters ical approach, like 3D viscous hydro + UrQMD with a for the fit function are listed in Table IV. For compari- consistent equation of state at non-zero baryon chemical son, the v2 from other energies are divided by the fit and potential. shown in the lower panels of Fig. 9. We choose 200 GeV data as the reference because the statistical errors are Figure 11 shows the energy dependence of smallest. For pT below 2 GeV c, the v2 values rise with v2 EtaSubs . Larger v2 EtaSubs values are ob- increasing collision energy. Beyond~ pT 2 GeV c the v2 served{ at higher} collision energy{ for} a selected p bin, results show comparable values within statistical= ~ errors. T but the pT dependence of the difference is quite different The increase of v2 pT as a function of energy could be from v2 4 . The ratios to 39 GeV data for each collision due to the change of( chemical) composition from low to energy{ first} decrease as a function of pT , then slightly high energies [22] and/or larger collectivity at the higher increase in the pT region of 1 - 2.5 GeV/c. The different collision energy. The baryonic chemical potential varies a trend of the energy dependence of v2 from v2 4 and lot (20 - 400 MeV) from 200 to 7.7 GeV [22]. The baryon v2 EtaSubs is interpreted as due to the different{ } over meson ratio is larger in lower collisions energies. The sensitivity{ of} the v2 methods to non-flow and/or flow difference of v2 for baryon and meson, for example proton fluctuations. 16

(a1) 7.7 GeV (b1) 11.5 GeV (c1) 39 GeV (d1) 200 GeV AMPTSM(3mb) AMPTSM(3mb) STAR data 0.2 0.2 Fitting lines 0.2 0.2 AMPTSM(10mb) AMPTSM(10mb) AMPT default AMPT default 0.15 UrQMD 0.15 0.15 UrQMD 0.15

{4} 0.1 0.1 0.1 0.1 2

v 0.05 0.05 0.05 0.05 0 0 0 0 0(a2) 0.5 Data/AMPT 1 default 1.5 2 (b2) Data/AMPTSM(3mb) (c2) (d2) 2.5 Data/UrQMD 2.5 Data/AMPTSM(10mb)2.5 2.5 2 2 2 2 1.5 1.5 1.5 1.5

Ratio 1 1 1 1 0.5 0.5 0.5 0.5 0 0.5 1 1.5 2 0 0.5 1 1.5 2 0 0.5 1 1.5 2 0 0.5 1 1.5 2 p (GeV/c) T

FIG. 12: (Color online) The v2 4 as a function of pT for 20 30% Au + Au collisions at √sNN = 7.7, 11.5, 39 and 200 GeV compared to corresponding results{ } from UrQMD, AMPT default− version, and AMPT with string melting version (3 and 10 mb). The shaded boxes show the systematic uncertainties for the experimental data of 7.7, 11.5 and 39 GeV. The bottom panels show the ratio of data to the ﬁt results of the models.

D. Model comparisons at all collision energies from 7.7 to 200 GeV. A larger parton cross section means stronger partonic interactions To investigate the partonic and hadronic contribution which translate into a larger magnitude of v2. The difference between data and these AMPT model calculations to the final v2 results from different collision energies, transport model calculations from AMPT default (ver. seems to show no significantly systematic change vs. col- 1.11), AMPT string-melting (ver. 2.11) [64] and UrQMD lision energies. However, a recent study with the AMPT (ver 2.3) [65] are compared with the new data presented. model suggests hadronic potentials affect the final v2 re- The initial-parameter settings for the models follow the sults significantly when the collision energy is less than recommendation in the cited references. The AMPT de- √sNN = 39 GeV [66]. fault and UrQMD models only take the hadronic interactions into consideration, while the AMPT string-melting version incorporates both partonic and hadronic interac- VI. SUMMARY tions. The larger the parton cross section, the later the hadron cascade starts. We have presented elliptic flow, v2, measurements from s Figure 12 shows the comparison of pT differential v2 4 Au + Au collisions at √ NN = 7.7, 11.5, 19.6, 27 and between model and data in the 20 − 30% centrality bin.{ } 39 GeV for inclusive charged hadrons at midrapidity. To The 200 GeV data are taken from [62]. The figure shows investigate non-flow correlations and v2 fluctuations, var- that UrQMD underpredicts the measurements at √sNN ious measurement methods have been used in the anal- = 39 and 200 GeV in the pT range studied. The differ- ysis. The difference between v2 2 and v2 4 decreases ences are reduced as the collision energy decreases. That with decrease in collision energy,{ indicating} { that} non-flow the ratio of data to UrQMD results are closer to 1 at the contribution and/or flow fluctuations decrease with a de- lower collision energy indicates that the contribution of crease in collision energy. The centrality and pT depen- hadronic interactions becomes more significant at lower dence of v2 are similar to that observed at higher RHIC collision energies. The AMPT model with default set- collision energies. A larger v2 is observed in more pe- tings underpredicts the 200 GeV data, while the ratios of ripheral collisions. The pseudorapidity dependence of v2 data to AMPT default results show no significant change indicates a change in shape from 200 GeV to 7.7 GeV from 7.7 to 39 GeV. The inconsistency between AMPT within the measured range −1 η 1, but the results of default and UrQMD makes the conclusion model depen- v2 versus pseudorapidity scaled< by< beam rapidity shows dent. The AMPT model with string-melting version with a similar trend for all collision energies. The compari- 3 and 10 mb parton cross sections overpredicts the results son with Au + Au collisions at higher energies at RHIC 17

s (√ NN = 62.4 and 200 GeV) and at LHC (Pb + Pb colli- tion of baryon and mesons vary significantly. s sions at √ NN = 2.76 TeV) shows the v2 4 values at low pT (pT 2.0 GeV/c) increase with increase{ } in collision energy implying< an increase of collectivity. The current viscous hydrodynamic simulations cannot reproduce the VII. ACKNOWLEDGMENTS trend of the energy dependence of v2 pT . The agreement between the data and( ) UrQMD, which is based on hadronic rescatterings, improves at lower We thank the RHIC Operations Group and RCF at collision energies, consistent with an increasing role of BNL, the NERSC Center at LBNL and the Open Science the hadronic stage at these energies. The inconsistency Grid consortium for providing resources and support. between AMPT default and UrQMD makes the conclu- This work was supported in part by the Offices of NP sion model dependent. The comparison to AMPT model and HEP within the U.S. DOE Office of Science, the U.S. calculations seems to show no significantly systematic NSF, the Sloan Foundation, CNRS/IN2P3, FAPESP change vs. collision energy, but improved calculations CNPq of Brazil, Ministry of Ed. and Sci. of the Russian including harmonic potentials may change the v2 values Federation, NNSFC, CAS, MoST, and MoE of China, from AMPT models at lower collision energies. GA and MSMT of the Czech Republic, FOM and NWO These results set the baseline to study the number of of the Netherlands, DAE, DST, and CSIR of India, Polish constituent quark scaling of identified hadron v2. It also Ministry of Sci. and Higher Ed., Korea Research Foun- sets the stage for understanding the collision energy de- dation, Ministry of Sci., Ed. and Sports of the Rep. of pendence of v2 in the regime where the relative contribu- Croatia, and RosAtom of Russia.

[1] S. A. Voloshin, A. M. Poskanzer and R. Snellings, 73, 044905 (2006); A.Andronic, P.Braun-Munzinger and arXiv:0809.2949 (2008). J. Stachel, Nucl. Phys. A 772, 167 (2006). [2] P. Sorensen, arXiv:0905.0174 (2009). [22] L. Kumar (For STAR Collaboration), J. Phys. G 38, [3] R. Snellings, New J. Phys. 13, 055008 (2011). 124145 (2011). [4] J. Y. Ollitrault, Phys. Rev. D 46, 229 (1992). [23] J. Cleymans et al., Phys. Rev. C 73, 034905 (2006). [5] C. Shen and U. Heinz, Phys. Rev. C 85, 054902 (2012). [24] Y. Aoki et al., Nature 443, 675 (2006). [6] K. H. Ackermann et al. (STAR Collaboration), Phys. [25] S. Ejiri, Phys. Rev. D 78, 074507 (2008). Rev. Lett. 86, 402 (2001). [26] M. A. Stephanov, Int. J. Mod. Phys. A 20, 4387 (2005). [7] J. Adams et al. (STAR Collaboration), Phys. Rev. Lett. [27] B. Mohanty, Nucl. Phys. A 830, 899c (2009) and refer- 92, 052302 (2004). ences therein. [8] J. Adams et al. (STAR Collaboration), Phys. Rev. Lett. [28] P.F. Kolb, J. Sollfrank, U. Heinz, Phys. Rev. C 62 (2000) 95, 122301 (2005). 054909. [9] J. Adams et al., (STAR Collaboration), Nucl. Phys. A [29] H. Sorge, Phys. Rev. Lett. 82, 2048 (1999). 757, 102 (2005). [30] A. M. Poskanzer and S. A. Voloshin, Phys. Rev. C 58 [10] J. Adams et al. (STAR Collaboration), Phys. Rev. C 72, 1671 (1998). 014904 (2005). [31] S. Wang et al., Phys. Rev. C 44, 1091 (1991). [11] B. I. Abelev et al. (STAR Collaboration), Phys. Rev. [32] N. Borghini, P. M. Dinh, and J.-Y. Ollitrault, Phys. Rev. Lett. 99, 112301 (2007). C 63, 054906 (2001). [12] B. I. Abelev et al. (STAR Collaboration), Phys. Rev. C [33] N. Borghini, P. M. Dinh, and J.-Y. Ollitrault, Phys. Rev. 77, 054901 (2008). C 64, 054901 (2001). [13] B. I. Abelev et al. (STAR Collaboration), Phys. Rev. C [34] R. S. Bhalerao, N. Borghini and J. Y. Ollitrault, Nucl. 81, 044902 (2010). Phys. A 727, 373 (2003). [14] S. S. Adler et al. (PHENIX Collaboration), Phys. Rev. [35] S. A. Voloshin, A. M. Poskanzer, A. Tang and G. Wang, Lett 91, 182301 (2003). Phys. Lett. B 659, 537 (2008). [15] S. S. Adler et al. (PHENIX Collaboration), Phys. Rev. [36] Md. Nasim et. al., Phys. Rev. C 82, 054908 (2010). Lett 94, 232302 (2005). [37] C. A. Whitten Jr. (For STAR Collaboration), AIP Conf. [16] B. Alver et al. (PHOBOS Collaboration), Phys. Rev. Lett Proc. 980, 390 (2008). 98, 242302 (2007). [38] C. Adler et al. (STAR Collaboration), Nucl. Instrum. [17] H. Appelshauser et al. (NA49 Collaboration), Phys. Rev. Methods A 470, 488 (2001). Lett 80, 4136 (998). [39] W. J. Llope et al. (STAR Collaboration), Nucl. Instrum. [18] K. Aamodt et al. (ALICE Collaboration), Phys. Rev. Methods A 522, 252 (2004). Lett. 105, 252302 (2010). [40] B. Bonner et al. (STAR Collaboration), Nucl. Instrum. [19] G. Aad et al. (ATLAS Collaboration), Phys. Lett. B 707, Methods A 508, 181 (2003); M. Shao et al. (STAR Col- 330 (2012). laboration), Nucl. Instrum. Methods A 558, 419 (2006). [20] B. Mohanty (For STAR Collaboration), J. Phys. G 38, [41] K. H. Ackermann et al. (STAR Collaboration), Nucl. In- 124023 (2011). strum. Methods A 499, 624 (2003). [21] J. Cleymans et al., Phys. Rev. C 73, 034905 (2006); F. [42] K. H. Ackermann et al. (STAR Collaboration), Nucl. In- Becattini, J. Manninen and M. Gazdzicki, Phys. Rev. C strum. Methods A 499, 713 (2003). 18

[43] B. I. Abelev et al. (STAR Collaboration), Phys. Rev. C [54] M. L. Miller, K. Reygers, S. J. Sanders, P. Steinberg, 81, 024911 (2010). Ann. Rev. Nucl. Part. Sci. 57, 205 (2007). [44] D. Kharzeev and M. Nardi, Phys. Lett. B 507, 121 [55] A. Adil et al., Phys. Rev. C 74, 044905 (2006). (2001). [56] H. J. Drescher and Y. Nara, Phys. Rev. C 75, 034905 [45] K. Nakamura et al. (Particle Data Group), J. Phys. G (2007). 37, 075021 (2010). [57] H. J. Drescher and Y. Nara, Phys. Rev. C 76, 041903 [46] B. B. Back et al. [PHOBOS Collaboration], Phys. Rev. (2007). C 70, 021902(R) (2004). [58] T. Hirano and Y. Nara, Phys. Rev. C 79, 064904 (2009). [47] J. Barrette et al. (E877 Collaboration), Phys. Rev. C 55, [59] S. Voloshin, nucl-th/0606022 (2006). 1420 (1997); I. Selyuzhenkov and S. Voloshin, Phys. Rev. [60] B. Alver et al. (PHOBOS Collaboration), Phys. Rev. C C 77, 034904 (2008). 77, 014906 (2008). [48] J. Barrette et al. (E877 Collaboration), Phys. Rev. C 56, [61] B. I. Abelev et al. (STAR Collaboration), Phys. Rev. C 3254 (1997). 75, 054906 (2007). [49] G. Agakishiev et al. (STAR Collaboration), Phys. Rev. [62] Y. Bai, Ph.D. thesis, Nikhef and Utrecht University, C 85, 014901 (2012). 2007. [50] A. Bilandzic, R. Snellings, S. Voloshin, Phys. Rev. C 83, [63] A. M. Poskanzer and S. A. Voloshin (NA49 Collabora- 044913 (2011). tion), Nucl. Phys. A 661, 341c (1999). [51] A. Bilandzic, PhD. thesis, Nikhef and Utrecht University, [64] Z. Lin et al. , Phys. Rev. C 72, 064901 (2005). 2011. [65] H. Petersen et al., arXiv: 0805.0567v1 (2008). [52] J. Y. Ollitrault, A. M. Poskanzer and S. A. Voloshin, [66] J. Xu et al., Phys. Rev. C 85, 041901 (2012). Phys. Rev. C 80, 014904 (2009). [53] M. Miller and R. Snellings, nucl-ex/0312008 (2003). 19

TABLE II: Summary of centrality bins, average number of participants Npart , number of binary collisions Ncoll , reaction ⟨ ⟩ ⟨ ⟩ plane eccentricity εRP , participant eccentricity εpart , root-mean-square the participant eccentricity εpart 2 and transverse ⟨ ⟩ ⟨ ⟩ { } area Spart from MC Glauber simulations at √sNN = 7.7, 11.5, 19.6, 27, and 39 GeV. The errors are systematic uncertainties. ⟨ ⟩ 2 Centrality (%) Npart Ncoll εRP εpart εpart 2 Spart (fm ) ⟨ ⟩ ⟨ ⟩ ⟨ ⟩ ⟨ ⟩ { } ⟨ ⟩ Au + Au at √sNN = 7.7 GeV 0-5% 337 2 774 28 0.043 0.007 0.102 0.003 0.117 0.003 25.5 0.4 ± ± ± ± ± ± 5-10% 290 6 629 20 0.10 0.01 0.14 0.01 0.16 0.01 23.0 0.3 ± ± ± ± ± ± 10-20% 226 8 450 22 0.18 0.02 0.21 0.02 0.24 0.02 19.5 0.4 ± ± ± ± ± ± 20-30% 160 10 283 24 0.26 0.03 0.30 0.02 0.32 0.02 15.7 0.7 ± ± ± ± ± ± 30-40% 110 11 171 23 0.32 0.04 0.37 0.03 0.39 0.03 12.6 0.8 ± ± ± ± ± ± 40-50% 72 10 96 19 0.36 0.04 0.43 0.03 0.46 0.03 10.0 0.9 ± ± ± ± ± ± 50-60% 45 9 52 13 0.39 0.04 0.50 0.03 0.53 0.03 7.8 1.0 ± ± ± ± ± ± 60-70% 26 7 25 9 0.40 0.05 0.58 0.04 0.62 0.04 5.8 1.1 ± ± ± ± ± ± 70-80% 14 4 12 5 0.36 0.05 0.68 0.04 0.72 0.04 3.6 1.0 ± ± ± ± ± ± Au + Au at √sNN = 11.5 GeV 0-5% 338 2 784 27 0.043 0.006 0.102 0.003 0.116 0.003 25.6 0.4 ± ± ± ± ± ± 5-10% 290 6 635 20 0.10 0.01 0.14 0.01 0.16 0.01 23.0 0.3 ± ± ± ± ± ± 10-20% 226 8 453 23 0.18 0.02 0.22 0.02 0.24 0.02 19.5 0.5 ± ± ± ± ± ± 20-30% 160 9 284 23 0.26 0.03 0.30 0.02 0.32 0.02 15.7 0.7 ± ± ± ± ± ± 30-40% 110 10 172 22 0.32 0.04 0.37 0.03 0.39 0.03 12.6 0.8 ± ± ± ± ± ± 40-50% 73 10 98 18 0.36 0.04 0.43 0.03 0.46 0.03 10.1 0.9 ± ± ± ± ± ± 50-60% 44 9 52 14 0.39 0.04 0.50 0.03 0.53 0.03 7.8 1.0 ± ± ± ± ± ± 60-70% 26 7 25 9 0.40 0.05 0.58 0.04 0.62 0.04 5.8 1.1 ± ± ± ± ± ± 70-80% 14 6 12 6 0.37 0.06 0.68 0.05 0.71 0.05 3.7 1.2 ± ± ± ± ± ± Au + Au at √sNN = 19.6 GeV 0-5% 338 2 800 27 0.044 0.006 0.102 0.003 0.117 0.003 25.6 0.4 ± ± ± ± ± ± 5-10% 289 6 643 20 0.11 0.01 0.15 0.01 0.16 0.01 23.0 0.3 ± ± ± ± ± ± 10-20% 225 9 458 24 0.18 0.02 0.22 0.02 0.24 0.02 19.5 0.5 ± ± ± ± ± ± 20-30% 158 10 284 26 0.26 0.03 0.30 0.02 0.32 0.02 15.6 0.7 ± ± ± ± ± ± 30-40% 108 10 170 23 0.32 0.04 0.37 0.03 0.40 0.03 12.5 0.8 ± ± ± ± ± ± 40-50% 71 10 96 18 0.36 0.04 0.43 0.03 0.46 0.03 10.0 0.9 ± ± ± ± ± ± 50-60% 44 9 51 13 0.39 0.04 0.50 0.03 0.53 0.03 7.8 1.0 ± ± ± ± ± ± 60-70% 25 7 25 8 0.40 0.05 0.58 0.04 0.62 0.04 5.8 1.1 ± ± ± ± ± ± 70-80% 14 5 12 5 0.37 0.06 0.68 0.05 0.71 0.05 3.7 1.2 ± ± ± ± ± ± Au + Au at √sNN = 27 GeV 0-5% 343 2 841 28 0.040 0.005 0.100 0.002 0.114 0.003 25.8 0.4 ± ± ± ± ± ± 5-10% 299 6 694 22 0.10 0.01 0.14 0.01 0.16 0.01 23.4 0.3 ± ± ± ± ± ± 10-20% 233 9 497 26 0.18 0.02 0.21 0.02 0.23 0.02 19.8 0.5 ± ± ± ± ± ± 20-30% 166 11 312 28 0.26 0.03 0.29 0.02 0.32 0.02 15.9 0.7 ± ± ± ± ± ± 30-40% 114 11 188 25 0.32 0.04 0.37 0.03 0.39 0.03 12.8 0.9 ± ± ± ± ± ± 40-50% 75 10 106 20 0.37 0.04 0.43 0.03 0.46 0.03 10.2 0.9 ± ± ± ± ± ± 50-60% 47 9 56 15 0.39 0.05 0.50 0.03 0.53 0.03 7.9 1.0 ± ± ± ± ± ± 60-70% 27 8 27 10 0.40 0.05 0.58 0.05 0.61 0.05 5.8 1.2 ± ± ± ± ± ± 70-80% 14 6 12 6 0.37 0.06 0.68 0.05 0.71 0.05 3.6 1.3 ± ± ± ± ± ± Au + Au at √sNN = 39 GeV 0-5% 342 2 853 27 0.042 0.006 0.101 0.003 0.115 0.003 25.9 0.4 ± ± ± ± ± ± 5-10% 294 6 687 21 0.10 0.01 0.14 0.01 0.16 0.01 23.3 0.3 ± ± ± ± ± ± 10-20% 230 9 492 26 0.18 0.02 0.21 0.02 0.23 0.02 19.8 0.5 ± ± ± ± ± ± 20-30% 162 10 306 27 0.26 0.03 0.30 0.02 0.32 0.02 16.0 0.7 ± ± ± ± ± ± 30-40% 111 11 183 24 0.32 0.04 0.37 0.03 0.39 0.03 12.8 0.8 ± ± ± ± ± ± 40-50% 74 10 104 20 0.36 0.04 0.43 0.03 0.46 0.03 10.3 1.0 ± ± ± ± ± ± 50-60% 46 9 55 14 0.39 0.04 0.50 0.03 0.53 0.03 8.0 1.0 ± ± ± ± ± ± 60-70% 26 7 27 9 0.40 0.04 0.58 0.04 0.61 0.04 5.9 1.1 ± ± ± ± ± ± 70-80% 14 5 12 6 0.37 0.05 0.67 0.05 0.71 0.05 3.8 1.2 ± ± ± ± ± ± 20

TABLE III: The εpart 2 and transverse area Spart from the Color Glass Condensate (CGC) model [55–58] calculations in { } ⟨ ⟩ Au + Au collisions at √sNN = 7.7, 11.5, 19.6, 27, 39, 62.4 and 200 GeV. The errors are systematic uncertainties. Centrality (%) 0 5% 5 10% 10 20% 20 30% 30 40% 40 50% 50 60% 60 70% 70 80% − − − − − − − − − Au + Au at √sNN = 7.7 GeV

εpart 2 0.104 0.005 0.19 0.01 0.29 0.01 0.39 0.02 0.47 0.02 0.54 0.03 0.59 0.03 0.62 0.03 0.51 0.02 { } 2 ± ± ± ± ± ± ± ± ± Spart (fm ) 25.9 1.3 21.8 1.1 17.5 0.9 13.4 0.7 10.2 0.5 7.7 0.4 5.5 0.3 3.6 0.2 1.8 0.1 ⟨ ⟩ ± ± ± ± ± ± ± ± ± Au + Au at √sNN = 11.5 GeV

εpart 2 0.104 0.005 0.19 0.01 0.29 0.01 0.39 0.02 0.47 0.02 0.53 0.03 0.59 0.03 0.62 0.03 0.51 0.02 { } 2 ± ± ± ± ± ± ± ± ± Spart (fm ) 25.2 1.2 21.2 1.1 17.0 0.9 13.0 0.7 9.9 0.5 7.5 0.4 5.4 0.3 3.5 0.2 1.8 0.1 ⟨ ⟩ ± ± ± ± ± ± ± ± ± Au + Au at √sNN = 19.6 GeV

εpart 2 0.105 0.005 0.19 0.01 0.29 0.01 0.39 0.02 0.47 0.02 0.53 0.03 0.58 0.03 0.61 0.03 0.51 0.02 { } 2 ± ± ± ± ± ± ± ± ± Spart (fm ) 24.4 1.2 20.6 1.0 16.6 0.9 12.6 0.7 9.7 0.5 7.3 0.4 5.3 0.3 3.5 0.2 1.8 0.1 ⟨ ⟩ ± ± ± ± ± ± ± ± ± Au + Au at √sNN = 27 GeV

εpart 2 0.105 0.005 0.19 0.01 0.29 0.01 0.39 0.02 0.47 0.02 0.53 0.03 0.58 0.03 0.61 0.03 0.51 0.02 { } 2 ± ± ± ± ± ± ± ± ± Spart (fm ) 24.1 1.2 20.3 1.0 16.4 0.8 12.5 0.6 9.6 0.5 7.2 0.4 5.3 0.3 3.5 0.2 1.8 0.1 ⟨ ⟩ ± ± ± ± ± ± ± ± ± Au + Au at √sNN = 39 GeV

εpart 2 0.105 0.005 0.19 0.01 0.29 0.01 0.39 0.02 0.47 0.02 0.53 0.03 0.58 0.03 0.61 0.03 0.50 0.02 { } 2 ± ± ± ± ± ± ± ± ± Spart (fm ) 23.9 1.2 20.1 1.0 16.2 0.8 12.4 0.6 9.5 0.5 7.2 0.4 5.3 0.3 3.5 0.2 1.8 0.1 ⟨ ⟩ ± ± ± ± ± ± ± ± ± Au + Au at √sNN = 62.4 GeV

εpart 2 0.105 0.005 0.19 0.01 0.29 0.01 0.39 0.02 0.47 0.02 0.53 0.03 0.58 0.03 0.61 0.03 0.50 0.02 { } 2 ± ± ± ± ± ± ± ± ± Spart (fm ) 23.7 1.2 20.0 1.0 16.1 0.8 12.3 0.6 9.4 0.5 7.2 0.4 5.3 0.3 3.5 0.2 1.8 0.1 ⟨ ⟩ ± ± ± ± ± ± ± ± ± Au + Au at √sNN = 200 GeV

εpart 2 0.104 0.005 0.19 0.01 0.29 0.01 0.39 0.02 0.47 0.02 0.53 0.03 0.57 0.03 0.60 0.03 0.49 0.02 { } 2 ± ± ± ± ± ± ± ± ± Spart (fm ) 23.7 1.2 20.0 1.0 16.1 0.8 12.3 0.6 9.4 0.5 7.2 0.4 5.3 0.3 3.6 0.2 1.9 0.1 ⟨ ⟩ ± ± ± ± ± ± ± ± ±

TABLE IV: Summary of the parameters for the ﬁt functions to the results of v2 4 vs. pT in Au+Au collisions at √sNN = 200 GeV. { }

Parameters p0 p1 p2 p3 p4 p5 10 20% 0.00730 0.00114 0.10785 0.00598 0.03941 0.01038 0.01508 0.00767 0.00411 0.00246 0.00041 0.00028 − − ± ± − ± ± − ± ± 20 30% 0.00890 0.00096 0.14250 0.00500 0.05206 0.00869 0.02156 0.00642 0.00685 0.00206 0.00077 0.00023 − − ± ± − ± ± − ± ± 30 40% 0.00581 0.00206 0.14526 0.01089 0.00529 0.01910 0.02409 0.01419 0.00797 0.00456 0.00084 0.00052 − − ± ± − ± − ± ± − ±