Disappearance of back-to-back high-pT hadron correlations in central Au+Au collisions at sqrt[s NN ] =200 GeV.

Abstract

Azimuthal correlations for large transverse momentum charged hadrons have been measured over a wide pseudorapidity range and full azimuth in Au+Au and p+p collisions at sqrt[s(NN)]=200 GeV. The small-angle correlations observed in p+p collisions and at all centralities of Au+Au collisions are characteristic of hard-scattering processes previously observed in high-energy collisions. A strong back-to-back correlation exists for p+p and peripheral Au+Au. In contrast, the back-to-back correlations are reduced considerably in the most central Au+Au collisions, indicating substantial interaction as the hard-scattered partons or their fragmentation products traverse the medium.

Full text

Page 1

arXiv:nucl-ex/0210033v1 25 Oct 2002
Disappearance of back-to-back high pT hadron correlations in central Au+Au
collisions at√sNN = 200 GeV
C. Adler11, Z. Ahammed23, C. Allgower12, J. Amonett14, B.D. Anderson14, M. Anderson5, G.S. Averichev9,
J. Balewski12, O. Barannikova9,23, L.S. Barnby14, J. Baudot13, S. Bekele20, V.V. Belaga9, R. Bellwied31,
J. Berger11, H. Bichsel30, A. Billmeier31, L.C. Bland2, C.O. Blyth3, B.E. Bonner24, A. Boucham26, A. Brandin18,
A. Bravar2, R.V. Cadman1, H. Caines33, M. Calder´ on de la Barca S´ anchez2, A. Cardenas23, J. Carroll15,
J. Castillo26, M. Castro31, D. Cebra5, P. Chaloupka20, S. Chattopadhyay31, Y. Chen6, S.P. Chernenko9,
M. Cherney8, A. Chikanian33, B. Choi28, W. Christie2, J.P. Coffin13, T.M. Cormier31, M.M. Corral16,
J.G. Cramer30, H.J. Crawford4, A.A. Derevschikov22, L. Didenko2, T. Dietel11, J.E. Draper5, V.B. Dunin9,
J.C. Dunlop33, V. Eckardt16, L.G. Efimov9, V. Emelianov18, J. Engelage4, G. Eppley24, B. Erazmus26, P. Fachini2,
V. Faine2, J. Faivre13, R. Fatemi12, K. Filimonov15, E. Finch33, Y. Fisyak2, D. Flierl11, K.J. Foley2, J. Fu15,32,
C.A. Gagliardi27, N. Gagunashvili9, J. Gans33, L. Gaudichet26, M. Germain13, F. Geurts24, V. Ghazikhanian6,
O. Grachov31, V. Grigoriev18, M. Guedon13, E. Gushin18, T.J. Hallman2, D. Hardtke15, J.W. Harris33,
T.W. Henry27, S. Heppelmann21, T. Herston23, B. Hippolyte13, A. Hirsch23, E. Hjort15, G.W. Hoffmann28,
M. Horsley33, H.Z. Huang6, T.J. Humanic20, G. Igo6, A. Ishihara28, Yu.I. Ivanshin10, P. Jacobs15, W.W. Jacobs12,
M. Janik29, I. Johnson15, P.G. Jones3, E.G. Judd4, M. Kaneta15, M. Kaplan7, D. Keane14, J. Kiryluk6,
A. Kisiel29, J. Klay15, S.R. Klein15, A. Klyachko12, T. Kollegger11, A.S. Konstantinov22, M. Kopytine14,
L. Kotchenda18, A.D. Kovalenko9, M. Kramer19, P. Kravtsov18, K. Krueger1, C. Kuhn13, A.I. Kulikov9,
G.J. Kunde33, C.L. Kunz7, R.Kh. Kutuev10, A.A. Kuznetsov9, L. Lakehal-Ayat26, M.A.C. Lamont3,
J.M. Landgraf2, S. Lange11, C.P. Lansdell28, B. Lasiuk33, F. Laue2, J. Lauret2, A. Lebedev2, R. Lednick´ y9,
V.M. Leontiev22, M.J. LeVine2, Q. Li31, S.J. Lindenbaum19, M.A. Lisa20, F. Liu32, L. Liu32, Z. Liu32, Q.J. Liu30,
T. Ljubicic2, W.J. Llope24, G. LoCurto16, H. Long6, R.S. Longacre2, M. Lopez-Noriega20, W.A. Love2, T. Ludlam2,
D. Lynn2, J. Ma6, D. Magestro20, R. Majka33, S. Margetis14, C. Markert33, L. Martin26, J. Marx15, H.S. Matis15,
Yu.A. Matulenko22, T.S. McShane8, F. Meissner15, Yu. Melnick22, A. Meschanin22, M. Messer2, M.L. Miller33,
Z. Milosevich7, N.G. Minaev22, J. Mitchell24, C.F. Moore28, V. Morozov15, M.M. de Moura31, M.G. Munhoz25,
J.M. Nelson3, P. Nevski2, V.A. Nikitin10, L.V. Nogach22, B. Norman14, S.B. Nurushev22, G. Odyniec15,
A. Ogawa21, V. Okorokov18, M. Oldenburg16, D. Olson15, G. Paic20, S.U. Pandey31, Y. Panebratsev9,
S.Y. Panitkin2, A.I. Pavlinov31, T. Pawlak29, V. Perevoztchikov2, W. Peryt29, V.A Petrov10, M. Planinic12,
J. Pluta29, N. Porile23, J. Porter2, A.M. Poskanzer15, E. Potrebenikova9, D. Prindle30, C. Pruneau31,
J. Putschke16, G. Rai15, G. Rakness12, O. Ravel26, R.L. Ray28, S.V. Razin9,12, D. Reichhold8, J.G. Reid30,
G. Renault26, F. Retiere15, A. Ridiger18, H.G. Ritter15, J.B. Roberts24, O.V. Rogachevski9, J.L. Romero5,
A. Rose31, C. Roy26, V. Rykov31, I. Sakrejda15, S. Salur33, J. Sandweiss33, I. Savin10, J. Schambach28,
R.P. Scharenberg23, N. Schmitz16, L.S. Schroeder15, A. Sch¨ uttauf16, K. Schweda15, J. Seger8, D. Seliverstov18,
P. Seyboth16, E. Shahaliev9, K.E. Shestermanov22, S.S. Shimanskii9, F. Simon16, G. Skoro9, N. Smirnov33,
R. Snellings15, P. Sorensen6, J. Sowinski12, H.M. Spinka1, B. Srivastava23, E.J. Stephenson12, R. Stock11,
A. Stolpovsky31, M. Strikhanov18, B. Stringfellow23, C. Struck11, A.A.P. Suaide31, E. Sugarbaker20, C. Suire2,
M.ˇSumbera20, B. Surrow2, T.J.M. Symons15, A. Szanto de Toledo25, P. Szarwas29, A. Tai6, J. Takahashi25,
A.H. Tang15, D. Thein6, J.H. Thomas15, M. Thompson3, V. Tikhomirov18, M. Tokarev9, M.B. Tonjes17,
T.A. Trainor30, S. Trentalange6, R.E. Tribble27, V. Trofimov18, O. Tsai6, T. Ullrich2, D.G. Underwood1,
G. Van Buren2, A.M. VanderMolen17, I.M. Vasilevski10, A.N. Vasiliev22, S.E. Vigdor12, S.A. Voloshin31,
F. Wang23, H. Ward28, J.W. Watson14, R. Wells20, G.D. Westfall17, C. Whitten Jr.6, H. Wieman15, R. Willson20,
S.W. Wissink12, R. Witt33, J. Wood6, N. Xu15, Z. Xu2, A.E. Yakutin22, E. Yamamoto15, J. Yang6, P. Yepes24,
V.I. Yurevich9, Y.V. Zanevski9, I. Zborovsk´ y9, H. Zhang33, W.M. Zhang14, R. Zoulkarneev10, A.N. Zubarev9
(STAR Collaboration)
1Argonne National Laboratory, Argonne, Illinois 60439
2Brookhaven National Laboratory, Upton, New York 11973
3University of Birmingham, Birmingham, United Kingdom
4University of California, Berkeley, California 94720
5University of California, Davis, California 95616
6University of California, Los Angeles, California 90095
7Carnegie Mellon University, Pittsburgh, Pennsylvania 15213
8Creighton University, Omaha, Nebraska 68178

Page 2

2
9Laboratory for High Energy (JINR), Dubna, Russia
10Particle Physics Laboratory (JINR), Dubna, Russia
11University of Frankfurt, Frankfurt, Germany
12Indiana University, Bloomington, Indiana 47408
13Institut de Recherches Subatomiques, Strasbourg, France
14Kent State University, Kent, Ohio 44242
15Lawrence Berkeley National Laboratory, Berkeley, California 94720
16Max-Planck-Institut fuer Physik, Munich, Germany
17Michigan State University, East Lansing, Michigan 48824
18Moscow Engineering Physics Institute, Moscow Russia
19City College of New York, New York City, New York 10031
20Ohio State University, Columbus, Ohio 43210
21Pennsylvania State University, University Park, Pennsylvania 16802
22Institute of High Energy Physics, Protvino, Russia
23Purdue University, West Lafayette, Indiana 47907
24Rice University, Houston, Texas 77251
25Universidade de Sao Paulo, Sao Paulo, Brazil
26SUBATECH, Nantes, France
27Texas A&M University, College Station, Texas 77843
28University of Texas, Austin, Texas 78712
29Warsaw University of Technology, Warsaw, Poland
30University of Washington, Seattle, Washington 98195
31Wayne State University, Detroit, Michigan 48201
32Institute of Particle Physics, CCNU (HZNU), Wuhan, 430079 China
33Yale University, New Haven, Connecticut 06520
Azimuthal correlations for large transverse momentum charged hadrons have been measured over
a wide pseudo-rapidity range and full azimuth in Au+Au and p+p collisions at√sNN = 200 GeV.
The small-angle correlations observed in p+p collisions and at all centralities of Au+Au collisions
are characteristic of hard-scattering processes already observed in elementary collisions. A strong
back-to-back correlation exists for p+p and peripheral Au + Au. In contrast, the back-to-back
correlations are reduced considerably in the most central Au+Au collisions, indicating substantial
interaction as the hard-scattered partons or their fragmentation products traverse the medium.
PACS numbers: 25.75
In collisions of heavy nuclei at high energies, a new
state of matter consisting of deconfined quarks and glu-
ons at high density is expected [1]. Large transverse mo-
mentum partons in the high-density system result from
the initial hard scattering of nucleon constituents. Af-
ter a hard scattering, the parton fragments to create a
high energy cluster (jet) of particles. A high momen-
tum parton traversing a dense colored medium is pre-
dicted to experience substantial energy loss [2, 3] and
may be absorbed. Measurement of the parton fragmen-
tation products after hard-scattering processes in nuclear
collisions may reveal effects due to the interaction of high-
momentum partons traversing the medium, thereby mea-
suring the gluon density of the medium [4].
Hard scattering processes have been established at high
transverse momentum (pT) in elementary collisions at
high energy through the measurement of jets [5, 6, 7],
correlated back-to-back jets (di-jets) [8], high pT single
particles, and back-to-back correlations between high pT
hadrons [9]. Jets have been shown to carry the momen-
tum of the parent parton [10]. The jet cross sections and
high pT single particle spectra are well described over
a broad range of energies [11] in terms of the hadron’s
parton distributions, hard parton scattering treated by
perturbative QCD, and subsequent fragmentation of the
parton. High pT jet events have also been studied in
proton-nucleus interactions [12]. In the absence of ef-
fects of the nuclear medium the rate of hard processes
should scale linearly with the number of binary nucleon-
nucleon collisions. Recent results from RHIC, however,
show a suppression of the single particle inclusive spectra
of hadrons for pT> 2 GeV/c in central Au+Au collisions,
indicating substantial in-medium interactions [13, 14].
In this Letter, we report measurements of two-hadron
angular correlations at large transverse momentum for
p+p and Au+Au collisions at√sNN= 200 GeV. These
correlation measurements provide the most direct ev-
idence for production of jets in high energy nucleus-
nucleus collisions, and allow for the first time measure-
ments, inaccessible in inclusive spectra, of the fate of
back-to-back jets in the dense medium as a function of
the size of the overlapping system. The results reveal

Page 3

3
significant interaction of hard-scattered partons (or their
fragmentation products) in the medium, with a strong
dependence on the geometry and distance of traversal.
The measurements were made using the STAR detec-
tor [15] at the Relativistic Heavy-Ion Collider (RHIC)
at Brookhaven National Laboratory. The STAR detec-
tor is a large acceptance magnetic spectrometer, with a
large volume Time Projection Chamber (TPC) inside a
0.5 Tesla solenoidal magnet. The TPC measures the tra-
jectories of charged particles and determines the particle
momenta. The TPC has full azimuthal coverage over
a pseudo-rapidity range |η| < 1.5. STAR has excellent
position and momentum resolution, and, due to its ver-
texing capabilities, is able to identify many sources of
secondary particles. The p+p analysis uses ≈10 million
minimum bias p+p events triggered on the coincidence
of signals from scintillator annuli spanning the pseudo-
rapidity interval 3.5 ≤ |η| ≤ 5.0. The Au+Au analysis
uses ≈1.7 million minimum bias Au+Au events and ≈1.5
million top 10% central Au+Au events.
Partons fragment into jets of hadrons in a cone around
the direction of the original hard-scattered parton. The
leading hadron in the jet tends to be most closely aligned
with the original parton direction [16]. The large multi-
plicities in Au+Au collisions make full jet reconstruction
impractical. Thus, we utilize two-particle azimuthal cor-
relations of high pT charged hadrons [17] to identify jets
on a statistical basis, with known sources of background
correlations subtracted.
Events with at least one large transverse momentum
hadron (4 < ptrig
T
< 6 or 3 < ptrig
fined to be a trigger particle, are used in this analysis.
For each of the trigger particles in the event, we incre-
ment the number N(∆φ,∆η) of associated tracks with
2 GeV/c < pT < ptrig
T
as a function of their azimuthal
(∆φ) and pseudo-rapidity (∆η) separations from the trig-
ger particle. We then construct an overall azimuthal pair
distribution per trigger particle,
T
< 4 GeV/c), de-
D(∆φ) ≡
1
Ntrigger
1
ǫ
?
d∆ηN(∆φ,∆η), (1)
where Ntriggeris the observed number of tracks satisfying
the trigger requirement. The efficiency ǫ for finding the
associated particle is evaluated by embedding simulated
tracks in real data. In order to have a high and con-
stant tracking efficiency, the tracks are required to have
|η| < 0.7, which translates to a relative pseudo-rapidity
acceptance of |∆η| < 1.4. The single track reconstruction
efficiency varies from 77% for the most central Au+Au
collisions to 90% for the most peripheral Au+Au colli-
sions and p+p collisions.
Identical analysis procedures are applied to the p+p
and Au+Au data.Displayed in Figure 1 are the az-
imuthal distributions for same-sign and opposite-sign
charged pairs from the a) p+p data and b) minimum
bias Au+Au data for 4 < ptrig
are integrated over the relative pseudo-rapidity range
0 < |∆η| < 1.4. Clear correlation peaks are observed
T
< 6 GeV/c. The data
a)
b)
c)
0
0. 1
0.4
0.5
0
0.1
1/Ntrigger dN/d(∆φ)
∆φ(radians)
-3 -2-1
012
3
same signopposite sign
0-10% Au+Au
Minimum Bias Au+Au
p+p
|∆η|<0.5 − |∆η|>0.5
FIG. 1: Azimuthal distributions of same-sign and opposite-
sign pairs for a) p+p, b) minimum bias Au+Au, and c)
background-subtracted central Au+Au collisions. All corre-
lation functions require a trigger particle with 4 < ptrig
GeV/c and associated particles with 2 GeV/c < pT < ptrig
The curves are one- or two- Gaussian fits.
T
< 6
T
.
near ∆φ ∼ 0 and ∆φ ∼ π in the data. The opposite-
sign correlations at small relative azimuth are larger than
those of the same-sign particle pairs, while the sign has
a negligible effect on the back-to-back correlations.
To isolate the jet-like correlations (localized in ∆φ,
∆η) in central Au+Au collisions, the azimuthal distri-
butions are measured for two regions of relative pseudo-
rapidity, |∆η| < 0.5 and 0.5 < |∆η| < 1.4 [17]. The
difference between the small and large relative pseudo-
rapidity azimuthal distributions is displayed in Figure 1c
along with single Gaussian fits. Near ∆φ = 0, the az-
imuthal distributions from Au+Au and p+p have simi-
lar shapes. For the opposite-sign azimuthal distributions,
the Gaussian widths are 0.17±0.01(stat.)±0.03(sys.) ra-
dians for p+p data, and 0.20 ± 0.02(stat.) ± 0.03(sys.)
radians for the central Au+Au data.
sign azimuthal distributions, the Gaussian widths are
0.16±0.02(stat.)±0.03(sys.) radians for p+p data, and
0.15 ± 0.03(stat.) ± 0.04(sys.) radians for the central
Au+Au data. The systematic errors reflect the spread
of values found for different choices of the ∆φ bin width.
Within the errors, there are no significant differences be-
tween the small-angle correlation widths for p+p and
central Au+Au collisions.
The ratios of the opposite-sign to same-sign peak ar-
eas are 2.7 ± 0.9(stat.) ± 0.2(sys.) for p+p and 2.5 ±
0.6(stat.)±0.2(sys.) for central Au+Au collisions. In jet
fragmentation, there are dynamical charge correlations
between the leading and next-to-leading charged hadrons
[18] that originate from the formation of q¯ q pairs along
a string between two partons. This results in a preferen-
tial ordering into oppositely-charged adjacent particles
along a string during fragmentation. The Hijing event
generator, which utilizes the Lund string fragmentation
scheme [19] incorporating these concepts, predicts a ratio
of 2.6±0.7 for the opposite-sign to same-sign correlation
For the same-

Page 4

4
strengths. The agreement of this ratio with those mea-
sured in the central Au+Au and p+p suggests that the
same jet production mechanism is responsible for a ma-
jority of the charged hadrons with pT > 4 GeV/c in p+p
and central Au+Au collisions.
The decay of resonances would also lead to small-angle
azimuthal correlations, but a resonance decay origin is
unlikely due to the observed correlation of particles with
the same charge sign, the similarity in the measured
small-angle azimuthal correlation widths in the Au+Au
and p+p interactions, and the strong back-to-back corre-
lations of large pTparticles seen for p+p collisions in Fig.
1a. The latter correlations, indicative of di-jet events [9],
are removed from the central Au+Au sample by the sub-
traction in Fig. 1c. A quantitative analysis of back-to-
back jet survival in Au+Au requires the more detailed
treatment of background correlations described below.
In addition to correlations due to jets, the two-particle
azimuthal distributions in Au+Au exhibit a structure at-
tributable to an anisotropy of single particle production
relative to the reaction plane. Previous measurements
[17] indicate that, even at large transverse momentum,
the particle distributions contain an anisotropy due to
elliptic flow that can be characterized by dN/d(φ−Φr) ∝
1+2v2cos(2(φ−Φr)), where Φris the reaction plane an-
gle determined event by event and v2is the elliptic flow
parameter. This leads to a two particle azimuthal distri-
bution of the form, dN/d∆φ = B(1+2v2
elliptic flow component of the two-particle azimuthal dis-
tribution is measured using several methods [17]. In this
paper, v2is determined using a reaction-plane method.
A simple reference model can be constructed for the
two-particle azimuthal distributions of high pT particles
in Au+Au collisions.A number of independent hard
scatterings (each similar to one measured in a triggered
p+p event) included in an event with correlations due to
elliptic flow can be represented by the azimuthal distri-
bution,
2cos(2∆φ)). The
Dmodel= Dpp+ B(1 + 2v2
2cos(2∆φ)).(2)
The elliptic flow parameter (v2) is measured indepen-
dently in the same set of events, and is taken to be con-
stant for pT > 2 GeV/c [17]. The parameter B is then
determined by fitting the observed DAuAuin the region
0.75 < |∆φ| < 2.24 radians, which is largely free of jet
contributions in the p+p data.
In Figure 2, the azimuthal distributions for 0 < |∆η| <
1.4 in Au+Au collisions at various centralities are com-
pared to Equation 2 using the measured p+p data. The
centrality selection is constructed by subdividing the
Au+Au minimum bias data sample into subsamples with
different charged particle multiplicities within |η| < 0.5.
The parameters v2and B are determined independently
for each centrality bin, and are listed in Table I. For
all centralities, the azimuthal correlation near ∆φ = 0
is well described by Equation 2. This indicates that the
same mechanism (hard parton scattering and fragmenta-
tion) is responsible for high transverse momentum parti-
-3 -2 -1
0
0.2
0.4
0.6
0.8
1
1.2
trigger
1/N
1.4
1.6
1.8
Au+Au data
p+p data + flow
))
φ∆
cos(2
2
2
B(1+2v
0-5%
10-20%
30-40%
60-80%
) φ
∆
dN/d(
(radians)
φ∆
0
1
23
FIG. 2: Azimuthal distributions (0 < |∆η| < 1.4, 4 < ptrig
6 GeV/c) for Au+Au collisions (solid circles) compared to the
expected distributions Dmodelfrom Equation 2 (open circles).
Also shown is the elliptic flow contribution for each centrality
(solid curve).
T
<
Centrality (%) Npart
60-80
40-60
30-40
20-30
10-20
5-10
0-5
v2
B
20±6
61±10 0.22±0.01 0.231± 0.003
114±13 0.21±0.01 0.420± 0.005
165±13 0.19±0.01 0.633± 0.005
232±11 0.15±0.01 0.931± 0.006
298±10 0.10±0.01 1.187± 0.008
352±7 0.07±0.01 1.442± 0.003
0.24±0.04 0.065± 0.003
TABLE I: Centrality, number of participants, v2 (2 < pT < 6
GeV/c), and normalization constant B.
and B are statistical only, while the errors on the number of
participants are systematic [14].
The errors on v2
cle production in p+p and Au+Au collisions. However,
the back-to-back correlations are suppressed in Au+Au
collisions compared to the expectation from Equation 2,
and the suppression is greater for more central collisions.
The most central collisions show no indication of any
back-to-back correlations beyond that expected from el-
liptic flow.
The ratio of the measured Au+Au correlation excess
relative to the p+p correlation is:
IAA(∆φ1,∆φ2) =
?∆φ2
∆φ1
d(∆φ)[DAuAu−B(1+2v2
2cos(2∆φ))]
?∆φ2
∆φ1
d(∆φ)Dpp
.(3)
The ratio can be plotted as a function of the number of
participating nucleons (Npart), deduced from the central-
ity bins as described in reference [14]. IAA is measured
for both the small-angle (|∆φ| < 0.75 radians) and back-
to-back (|∆φ| > 2.24 radians) regions. The ratio should

Page 5

5
Npart
100 150 200 250 300 350
0
0.5
1
1.5
2

100 150 200 250 300 350 400
400
0 50
50
|∆φ|<0.75
|∆φ|>2.24
IAA(∆φ1,∆φ2)
Npart
4<pT <6 GeV/c
trig
3<pT <4 GeV/c
trig
FIG. 3: Ratio of Au+Au and p+p (Equation 3) for small-
angle (squares, |∆φ| < 0.75 radians) and back-to-back (cir-
cles, |∆φ| > 2.24 radians) azimuthal regions versus num-
ber of participating nucleons for trigger particle intervals
4 < ptrig
T
< 6 GeV/c (solid) and 3 < ptrig
low). The horizontal bars indicate the dominant systematic
error (highly correlated among points) due to the uncertainty
in v2.
T
< 4 GeV/c (hol-
be unity if the hard-scattering component of Au+Au col-
lisions is simply a superposition of p+p collisions unaf-
fected by the nuclear medium. These ratios are given
in Figure 3 for the trigger particle momentum ranges
indicated. The asymmetric systematic errors are domi-
nated by the +5/-20% systematic uncertainty on v2due
to the potential non-flow contributions [20] as well as
other sources of systematic uncertainty [17].
For the most peripheral bin (smallest Npart), both
the small-angle and back-to-back correlation strengths
are suppressed compared to the expectation from Equa-
tion 2. This may be an indication of initial state nu-
clear effects such as shadowing of parton distributions
or scattering by multiple nucleons, or may be indica-
tive of energy loss in a dilute medium [21]. As Npart
increases, the small-angle correlation strength increases,
with a more pronounced increase for the trigger parti-
cles with lower pT threshold. If there were a large non-
jet contribution to particle production (i.e.
transverse flow) at the trigger threshold and above, it
would dilute the jet related correlation signal and this
ratio would be reduced. The back-to-back correlation
strength, above background from elliptic flow, decreases
with increasing Npartand is consistent with zero for the
most central collisions.In the extreme case, if there
were no elliptic flow for the 0-5% most central collisions,
IAA(2.24,π) = 0.4 ± 0.1 for 4 < ptrig
pared to IAA(2.24,π) = 0.1 ± 0.1 using the measured
elliptic flow value. Therefore, an overestimation of the
elliptic flow cannot explain the observed suppression of
back-to-back correlations.
Analyses of fixed-target experiments [22] have sug-
gested that the shape of the back-to-back dihadron az-
imuthal distribution is sensitive to the intrinsic parton
transverse momentum kT in the initial state. In proton-
collective
T
< 6 GeV/c, com-
nucleus and nucleus-nucleus collisions, additional initial-
state transverse momentum can be generated by multi-
ple nucleon-nucleon interactions preceding a hard scat-
tering [23, 24, 25]. To investigate whether this nuclear
kT can account for the observed deficit of back-to-back
azimuthal correlations in central Au+Au collisions, we
have carried out Pythia [19] simulations varying the kT
parameter. A rather extreme change from the nominal
value of σ = 1 GeV/c to 4 GeV/c introduced only a
small effect, reducing the predicted IAA(2.24,π) by less
than 20%. Experimental study of initial state effects on
the azimuthal correlations requires the measurement of
p(d)+Au collisions at RHIC energies.
In addition to the present data, two other striking
effects have been observed at high pT in nuclear colli-
sions at RHIC: strong suppression of the inclusive hadron
yield in central collisions [13, 14], and large elliptic flow
which saturates at pT > 3 GeV/c [17]. These phenom-
ena are all consistent with a picture in which observed
hadrons at pT > 3−4 GeV/c are fragments of hard scat-
tered partons, and partons or their fragments are strongly
scattered or absorbed in the nuclear medium. The ob-
served hadrons therefore result preferentially from hard-
scattered partons generated on the periphery of the reac-
tion zone and heading outwards [26]. In this picture the
inclusive yield will be suppressed relative to the binary
scaling expectation, and the strong position-momentum
correlation required to explain the large elliptic flow [27]
emerges naturally. The properties of small-angle hadron
correlations will have weak dependence on the size of the
colliding system, whereas the back-to-back correlations
will exhibit strong suppression for a large system relative
to a small one, both as observed.
In summary, STAR has measured azimuthal correla-
tions for high pT charged particles over a large relative
pseudo-rapidity range with full azimuthal angle coverage.
Comparison of the opposite-sign and same-sign correla-
tion strengths indicates that hard scattering and frag-
mentation is the predominant source of charged hadrons
with pT > 4 GeV/c in central Au + Au collisions. The
azimuthal correlations in Au+Au collisions have been
treated as the superposition of independently determined
elliptic flow and individual hard parton scattering con-
tributions, the latter measured in the STAR p+p data.
The most striking feature of the hard-scattering com-
ponent is an increasing suppression of back-to-back rela-
tive to small-angle correlations with increasing centrality.
These observations appear consistent with large energy
loss in a system that is opaque to the propagation of high-
momentum partons or their fragmentation products.
We wish to thank the RHIC Operations Group and the
RHIC Computing Facility at Brookhaven National Labo-
ratory, and the National Energy Research Scientific Com-
puting Center at Lawrence Berkeley National Laboratory
for their support. This work was supported by the Divi-
sion of Nuclear Physics and the Division of High Energy
Physics of the Office of Science of the U.S. Department
of Energy, the United States National Science Founda-

Page 6

6
tion, the Bundesministerium fuer Bildung und Forschung
of Germany, the Institut National de la Physique Nu-
cleaire et de la Physique des Particules of France, the
United Kingdom Engineering and Physical Sciences Re-
search Council, Fundacao de Amparo a Pesquisa do Es-
tado de Sao Paulo, Brazil, the Russian Ministry of Sci-
ence and Technology and the Ministry of Education of
China and the National Natural Science Foundation of
China.
[1] See forexample: J.W.Harrisand B.M¨ uller,
Annu. Rev. Nucl. Part. Sci. 46, 71 (1996).
[2] M. Gyulassy and M. Pl¨ umer, Phys. Lett. B243, 432
(1990).
[3] X. N. Wang and M. Gyulassy, Phys. Rev. Lett. 68, 1480
(1992).
[4] See for example: R. Baier, D. Schiff, and B. G. Zakharov,
Annu. Rev. Nucl. Part. Sci. 50, 37 (2000).
[5] M. Banner et al., Phys. Lett. B118, 203 (1982).
[6] G. Arnison et al., Phys. Lett. B123, 115 (1983).
[7] F. Abe et al., Phys. Rev. Lett. 62, 613 (1989).
[8] F. Abe et al., Phys. Rev. D41, 1722 (1990).
[9] G. Arnison et al., Phys. Lett. B118, 173 (1982).
[10] F. Abe et al., Phys. Rev. Lett. 65, 968 (1990).
[11] T. Akesson et al., Phys. Lett. B123, 133 (1983); J.A.
Appel et al., Phys. Lett. B160, 349 (1985); G. Arnison et
al., Phys. Lett. B172, 461 (1986); J.F. Owens, Rev. Mod.
Phys. 59, 465 (1987).
[12] B. Brown et al., Phys. Rev. Lett. 50, 11 (1983); R. Gomez
et al., Phys. Rev. D35, 2736 (1987); H. Miettinen et al.,
Phys. Lett. B207, 222 (1988); R.C. Moore et al., Phys.
Lett. B244, 347 (1990).
[13] K. Adcox et al., Phys. Rev. Lett. 88, 022301 (2002).
[14] C. Adler et al., nucl-ex 0206011.
[15] C. Adler et al., submitted to Nucl. Inst. Meth.
[16] G. Arnison et al., Phys. Lett. B132, 223 (1983).
[17] C. Adler et al., nucl-ex/0206006.
[18] P. Abreu et al., Phys. Lett. B407, 174 (1997).
[19] T. Sj¨ ostrand, P. Ed´ en, C. Friberg, L. L¨ onnblad, G. Miu,
S. Mrenna and E. Norrbin, Comp. Phys. Commun. 135,
238 (2001).
[20] C. Adler et al., Phys. Rev. C66, 034904 (2002).
[21] F. Arleo, Phys. Lett. B532, 231 (2002); E. Wang and
X.N. Wang, Phys. Rev. Lett. 89, 162301 (2002).
[22] L. Apanasevich et al., Phys. Rev. Lett. 81, 2642 (1998).
[23] J.W. Cronin et al., Phys. Rev. D11, 3105 (1975).
[24] M.D. Corcoran et al., Phys. Lett. B259, 209 (1991).
[25] M. Lev and B. Petersson, Z. Phys. C21, 155 (1983).
[26] J.D. Bjorken, Fermilab-Pub-82/59-THY (1982).
[27] E.V. Shuryak, Phys. Rev. C66, 027902 (2002).