The question we are trying to answer since 20 years:

1
Physics objective of charmonium production study
in heavy ion collisions
The question we are trying to answer since 20
years
Is there any evidence of J/Y suppression due to
the formation of a QGP in nucleus-nucleus
collisions ?

• Before considering this question, one should try
  to consider in detail if the J/Y production can
  be affected also by a cold nuclear environment,
  as in pA collisions
• Notice that here we dont use data to make
  studies of the spectral shape. The sought effect
  changes the yield, and that is what is measured
  
• Study carried out by NA38/NA50/NA60 at the SPS
  both in pA (400, 450 GeV/c) and in In-In and
  Pb-Pb (158 GEV/c) from 1986 until today
  
• Very significant contributions in p-A (800 and
  920 GeV/c) by E866 and HERA-B
  
• Important contribution starting to come also from
  PHENIX with data on d-Au, Au-Au and Cu-Cu (200
  GeV/c)

2
J/? (and Drell-Yan) production in proton-nucleus
(p-A) collisions
J/? (and DY) is produced in hard scattering
processes in some nucleon-nucleon collision.
The production cross-section in pA can be relate
d to the cross section in pp by
Where rA is the nuclear density distribution.
Since it is normalized to A nucleons, we get
Thus we expect the pA production cross section to
scale linearly with the number of nucleons.
DY follows this scaling law to a rather high deg
ree of precision. However, this is not observed
for J/? and other charmonia states ...
3
J/? and ? absorption in p-nucleus collisions at
the SPS
The J/y and y production cross-sections scale
less than linearly with the number of target
nucleons (contrary to what happens with high-mass
Drell-Yan dimuons).
NA50 p-A data collected in year 2000,with Be,
Al, Cu, Ag, W and Pb targets
If A dependence can be parametrized as
typical values of a are between 0.9 and 1 (1no
suppression)
Note the spA spp x Aa parameterization leads
to extrapolated spp(J/y) and spp(y) values which
are 10 to 20 higher than those obtained using
the Glauber model
4
The non linear increase as a function of A can be
related to an absorption cross-section
Where Sabs is the surviving probability of a J/?
produced in z
This is strictly correct in the hypothesis that
suppression is only due to dissociation by
nucleons. However, as we will see, other effects
can be responsible for the suppression and this
can bias the nuclear absorption cross-section.
A good approximation of the surviving probabilit
y is
L is the effective path length which the J/y
and y states traverse in the target nucleus,
from the production point of the ccbar pair to
the nuclear surface
5
SPS energy J/? in p-p and p-A collisions at 400
and 450 GeV/c

• Calculation of ?abs using the Glauber model
• Several data sets (collected in 10 years)
• Results on
• Absolute cross section
• Cross section ratios (J/?/DY)
• ?abs determination looks robust

6
What is ?absJ/? ?

• The ?absJ/? extracted from pA data is in reality
  an effective quantity, call it ?absJ/?,eff
  
• First, the J/? we observed are the sum of
  directly produced J/? and feed-down from higher
  mass states (? and ?c). The surviving
  probability should be written in reality as

(there is not a feed-down problem for the ? and
?c) Only if ?absJ/? , ?abs? , ?abs? then the
feed-down does not affect the absorption.
Are ?absJ/? , ?abs? , ?abs? equal or different?
Dipends on what state crosses the nuclear matter
? pre-resonant or fully formed state?
In principle one should measure the A dependence
of the y and cc
?important to understand in detail the anomalous
suppression seen in A-A collisions
In addition, there are initial state effects,
besides (final state) absorption that can lead to
suppression. This are just ignored in the SPS
measurements, wrapped up into ?absJ/?,eff
7
Other charmonium states ?

• Studied in detail by E866, NA50 and, more
  recently, by HERA-B
  
• Around xF 0, ?absJ/? ? ?abs? (partial
  hadronization ?)
  
• ?abs? (7.9 ? 0.6) mb (NA50) to be compared with
  ?abseff,J/? (4.1 ? 0.4) mb
  
• ? more loosely bound ? should be much more
  suppressed than J/?
  
• ?? 0.858 ? 0.017 ? 0.008 (NA50) ?
  significantly smaller than E866 value (0.92) ?
  energy dependence ?

8
?c production at HERA B (pA_at_920 GeV)
entries/(10 MeV/c2)
Fraction of the J/y yieldresulting from cc
decays
cc
From the 2000 data sample 370 74 ccs (mm e
e-) R(?c) 0.32 0.06 0.04 Phys. Lett. B
561 (2003) 61
2002/2003 data 40 times larger ?c statistics
9
Feed-down from the cc
21 5 of the J/y mesons observed by HERA-B are
due to cc decays Lower than the 3040 values com
ing from earlier data (inc. 2000 HERA-B data)
Based on 1300 ccs reconstructed in the dimuon
channel
Of the observed J/y mesons 7 are from y
decays, 20 from cc decays? more than 70 are
directly produced
10
Shadowing
Measurements of the nuclear charged parton
distributions by deep-inelastic scattering off
both a large nuclear target and a deuterium
target show that the ratio RF2F2A/F2D has a
characteristic shape as a function of x.
The region below x0.1 is referred as the

shadowing region The range 0.37 is know as the EMC

region In both regions the parton density is de
pleted
in the heavy
nucleus relative to the deuteron

(RF2 saturate Between the shadowing and the EMC reg
ions,
an enhancement,
antishadowing is seen, where RF2 1.
There is also an enhancement as x?1 assumed to b
e due to nucleon Fermi motion.
11
Nuclear effects on PDFs and absorption vs A at
the SPS
Calculations and figuremade by Ramona Vogt
sabs 0 mb
sabs 4 mb
sabs 7 mb
EKS98 0
With anti-shadowing on the PDFs, the J/y
production cross-section per nucleon increases
from pp to p-Pb, if there is no final state
absorption If some level of anti-shadowing is p
resent, the effective absorption cross section
would be actually underestimated (important if we
want to compare to RHIC data, where the effects
of shadowing are included in the nuclear matter
effects)
12
Setting the baseline at RHIC dAu collisions
At RHIC energies, for charm production, the
nuclear effects on the parton densities
(according to EKS98) are just in the crossing
from anti-shadowing to shadowing, and have a
significant impact on the rapidity dependence of
the measured absorption.

• Clear asymmetry observed in the rapidity
  dependence
  
• y0 small x2 (0.003) ? shadowing
• y

13
Low x2 0.003 (shadowing region)
Klein,Vogt, PRL 91142301,2003
(x2 is x in the nucleus)
Data favors weak shadowing absorption
How much is shadowing? No x2 scaling
How much is nuclear absorption?
? Need more data
Total effect possibly weaker than at fixed target
energy
14
Charmonia nuclear absorption vs xF
a strongly decreases at high xF ... Why is this
so? Higher parton densities? If so, the J/y shou
ld be strongly absorbed in d-Au at RHIC energies
and it is not...

• Cold nuclear effects at all xF
• Nuclear effects on the PDFs
• Final state ccbar absorption
• Cold nuclear effects at forward xF
• Energy loss of the incident parton
• Intrinsic charm in the nucleons

PHENIX coverage
Only a combination of absorption and shadowing
effects may be sufficient to describe the J/Y data
NA38/NA50/NA60 coverage
15
What about negative xF ?

• Region where the fully-formed resonance should
  interact with the
  
• nuclear medium

? First recent measurements by HERA-B
15 of full sample

• Preliminary results show a rather flat behaviour
  in the negative xF region

16
Cold Nuclear Matter Transverse Momentum
Broadening
Initial-state gluon multiple scattering causes pT
broadening (or Cronin effect)
Increase of ? with pT commonly understood as
parton scattering in the initial state (E866 data)
17
High x2 0.09
NA50
PHENIX 200 GeV results show pT broadening
comparable to that at lower energy (?s39 GeV in
E866/NuSea)
Low x2 0.003
The increase of a with pT seems to be identical
at 400, 800, 920 GeV and at RHIC
(at mid-rapidity) ? Maybe the increase of a from
NA50 to E866 to HERA-B to PHENIX is due to
the increase of the average pT of the J/y when ?s
increases...
18
J/? absorption in cold nuclear matter where are
we ?
Accurate p-A data exist, at least for J/?
- 400/450 GeV, in a restricted xF domain
- 800 GeV, wide xF domain - 920 GeV data, xF-0.3
, final results expected soon - PHENIX dAu data (
but statistics not very high) - Reference for J/?
suppression in A-A collisions established (need
to be improved for PHENIX
Theoretical description is not straightforward.
In principle what seen in experimental data is
the convolution of several effects
- Nuclear PDFs


- Energy loss in the initial state

- Nuclear absorption


- Energy loss in the final state

- Intrinsic charm
However, at SPS and RHIC (so far) only Nuclear P
DFs and final state absorption is considered
(important)
19

• A complete calculation of nuclear matter effects
  would require the knowledge of feed-down from
  higher mass resonances (y and cc) and their A
  dependence
• ? knowledge is fairly good (may be improved)
• ?c knowledge is quite poor
• -Data on A-dependence to appear soon by HERA-B
• -To be followed (hopefully) by NA60 at 400 GeV
• - cc at PHENIX

• Other points
• Complete xF, pT dependence of charmonia
  production in p-A exists
  
• only at one incident energy (E866)
• Comparison NA50/E866/PHENIX suggests a possible
  energy dependence
  
• of the various nuclear effects
• ?Complete the analysis of the NA60 pA data _at_
  158 GeV/c

20
J/y suppression and QGP formation
In 1986, Matsui and Satz argued that the
suppression of the J/y production yield in
nuclear collisions should be a clear signal of
the phase transition from confined hadronic
matter to a deconfined plasma of quarks and
gluons.
more than 975 citations!

• ...we thus conclude that
• there appears to be no mechanism for J/y
  suppression in a nuclear collision except the
  formation of a plasma
  
• and if such a plasma is produced, there seems to
  be no way to avoid J/y suppression

20 years ago!
21
In a deconfined phase the QCD bindingpotential
is Debye screened and theheavy quarkonia states
are dissolved.In other words, the free hard
gluons areenergetic enough to break the bound
QQstates into open charm and beauty mesons.
The formation of a QCD medium with deconfined
quarks and gluons will very significantly affect
the centrality dependence of quarkonia production
yields
22
Different heavy quarkonium states have different
binding energies and, hence, are dissolved at
successive thresholds in energy density or
temperature of the medium
A smoking gunsignal of the QGP
Feed-down from higher states ? step-wise J/y and
? suppression patterns? a thermometer of the
produced QCD matter
23
J/? absorption in hot nuclear matter
To assess whether there is further suppression in
hot nuclear matter, we measure the rate and
compare to the expected due only to the presence
of cold nuclear matter. The absorption cross se
ction found from pA data can be used in a Glauber
analysis of A-A data to determine the expected
production rate (ds/dy)G because of cold nuclear
matter effects. We can define a surviving proba
bility
What can we say if SJ/Yhadronic mechanisms which can suppress further
the J/Y with respect to pA?
Absorption by comovers no threshold
Already present in light ion collisions (even if
the effect might be small because
of the lower comover density)
24

• Take all the existing p-A data in the SPS energy
  domain
  
• NA38/NA50 400/450 GeV
• ? constrain ?absJ/?
• NA3/NA38 200 GeV
• ? constraint ?ppJ/?
• good compatibility with the
• S-U data at 200 GeV
• Shows that the hadronic comovers
• have no sizeable effect on the J/?

But notice that ?absJ/?,SU (7.2 ? 3.2) mb
To decrease the error a very large
statistics would be needed (short lever arm of t
he Glauber fit)
25
Pb-Pb data (NA50) The anomalous J/? suppression
Previous data sets partly biased

• Both with the proposed reference.

This is the more recent (and the last) set of dat
a
from NA50

• The central points show a departure
• from the normal absorption reference

26
The anomalous J/? suppression In-In data (NA60)

• What can we learn by comparing the J/?
• suppression pattern between different systems?

Which is the variable governing the onset of
the anomalous suppression? Discriminate between
models!

• Scaling behavior between various systems for a
  given centrality
  
• variable may indicate that that centrality
  variable is behind the
  
• observed anomalous suppression

• For instance, for L 7 fm, S-U, In-In and Pb-Pb
  collisions probe different values of Npart,
  ranging from 80 to 130
  
• ?If the physics-driving variable is Npart, the
  three systems will show a different pattern

27
J/? / DY vs. centrality
Anomalous suppression present in Indium-Indium

• Qualitative agreement with
• NA50 results plotted as a
• function of Npart

• Data points have been normalized to the
  expected J/? normal nuclear
  
• absorption, calculated with
• as measured with p-A NA50 data
• at 400 and 450 GeV

?J/?abs 4.18 ? 0.35 mb
B. Alessandro et al., Eur. Phys. J. C39(2005) 335
bin1 ? ?Npart? 63 bin2 ? ?Npart? 123 bin3 ?
?Npart? 175
3 centrality bins, defined through EZDC
28
J/? yield vs nuclear absorption

• Compare data to the expected J/? centrality
  distribution, calculated
  
• assuming nuclear absorption (with ?abs 4.18
  mb) as the only
  
• suppression source

Nuclear absorption
require the ratio measured/expected, integrated
over centrality, to be equal to the same quantity
from the (J/?)/DY analysis (0.87 0.05)
Normalization of the nuclear absorption curve
29
Small statistical errors Careful study of syste
matic
errors is needed

• Uncertainty on normal nuclear absorption
  parameters (?abs(J/?) and ?pp(J/?))
  
• Uncertainty on relative normalization between
  data and absorption curve
  
• Uncertainty on centrality determination
• Glauber model parameters
• EZDC to Npart

• 10 error centrality indep.? does not affect
  shape of the distribution
  
• Partly common to analysis a and b
• (Most) Central points affected by a considerable
  error

30
In-In and Pb-Pb vs Npart
NA50 Npart estimated through ET (left), or EZDC
(right, as in NA60)

• Good agreement with PbPb
• S-U data show a different behavior (asymmetric
  system ?)

31
In-In and Pb-Pb vs energy density and fireball
transverse size

• Suppression vs energy density
• Anomalous suppression sets in
• at ?1 GeV/fm3

• Suppression vs transverse size
• of the fireball
• Scaling variable for the onset ?

Quantitative comparison would need Pb-Pb data
with smaller error bars
32
Threshold effect or smooth onset (In-In)?
Step position Npart 86 8 ( ?Bj 1.6 GeV/fm
3 ) A1 0.98 0.02 A2 0.84 0.01 ?2/dof 0.
7

• Taking into account centrality smearing, data
  are in agreement with a sharp drop
  
• A smoother onset can not anyway be excluded

33
J/? suppression at fixed target intermediate
conclusions (only experimental)

• Although we are dealing with a hard process,
  quantitative calculations of the production and
  hadronic suppression are difficult
  
• Fortunately, very good quality data exist from
  p-p to Pb-Pb collisions
  
• (NA50, E866, HERA-B)
• Baseline for the interpretation of A-A data
• High statistics A-A data
• Anomalous suppression seen both in Pb-Pb (NA50)
  and in In-In (NA60)
  
• Physics interpretation still evolving but
  lets see first the RHIC data
  
• An important reference has been set for higher
  energy
  
• experiments at RHIC and LHC

34
Extend the picture charmonium production at RHIC
PHENIX is starting to play a serious role in J/Y
suppression RUN 4 1000 J/Y for ? ? 0.35
4450 J/Y for 1.2? ? ? 2.4

?mee105 MeV

• PHENIX experiment
• e? (? ? 0.35, p ? 0.2 GeV/c)
• ?? (1.2
• Larger kinematical domain with respect
• to the SPS
• Simultaneous measurement of e and ?

?m??165 MeV
35
Charmonium at RHIC questions

• The basic question addressed is the same we have
  to answer at SPS
  
• New production mechanisms
• Ncc (RHIC) 10
• May lead to enhancement from cc recombination as
  the collision
  
• volume cools
• New backgrounds
• Feed-down from B-decays
• The baseline should be studied as accurately as
  at SPS
  
• d-Au vs p-p to understand shadowing/absorption
• The nuclear suppression (A-A) should be studied
  for many systems
  
• to help disentangling the effects not connected
  with deconfinement
  

36
8
RAA vs Npart in AuAu collisions
RAA y yield divided by Ncoll and scaled to the
pp value
RAA
1

• RAA vs. Npart. (pT integrated) for
• y
• 1.2

Bar uncorrelated error Bracket correlated erro
r
0

• Different behavior in RAA
• between mid-rapidity and
• forward-rapidity.
• J/Y suppression is larger
• at forward-rapidity than
• at mid-rapidity
• S 0.6 for Npart100

1
RAAforward/RAAmid
S RAA (1.2
0
37
9
RAA and cold nuclear matter (CNM) effects

• CNM effects
• Gluon shadowing
• nuclear absorption
• J/y measurement in dAu collisions.
• sabs 1mb
• PRL, 96, 012304 (2006)

RAA
1
RHIC CNM effects (sabs 0, 1, 2mb at y0, y2)
R. Vogt et al., nucl-th/0507027
0
Significant suppression relative to CNM effects.
CNM effects predict larger suppression at
mid-rapidity, while data shows larger suppression
at forward-rapidity.
38
J/y suppression pattern SPS vs. RHIC data
RHIC y yield divided by Ncoll and scaled to the
pp value S-U and Pb-Pb y/DY ratio scaled to th
e pp value (equivalent since the DY cross section
scales linearly with Ncoll) In-In y yield divi
ded by the y nuclear absorption and multiplied by
the y/DY nuclear absorption (equivalent to
measured y over expected DY) and scaled to the
pp value This RAA pattern includes the nuclear
absorption and the extra suppression
(y/DY)AA / (y/DY)pp
(y/Ncoll)AA / (y/Ncoll)pp
Surprisingly good overlap between SPS and RHIC
J/y suppression patterns, at mid-rapidity, vs.
Npart
39
Comparison with theoretical models comovers
The new In-In data and RHIC tell a lot, when
compared to theoretical predictions that were
tuned to reproduce the Na50 data Pb-Pb
Suppression by hadron comovers
(Capella-Ferreiro) nuclear absorption and comover
s interaction sabs 4.5 mb sco 0.65 mb (C
apella-Ferreiro) Tuned on NA50 data
NA60 In-In _at_ 158 GeV
The prediction of Capella and Ferreiro fails to
describe the In-In data
Pb-Pb _at_ 158 GeV
40
Dissociation by comoving partons and hadrons at
RHIC
18
Capella et al., hep-ph/0610313
Calculation for y0 and y1.8
Data shows opposite trend
41
Models with dissociation and recombination
In a central Au-Au collision at RHIC more than cc
pairs can be produced.
? It becomes likely to have recombination
of uncorrelated pairs in the QGP to form J/Y
Models were developed where initially produced J
/Y and (continuous) regeneration in a QGP are
both taken into account simoultaneously.
Yield determined by the balance of the two compe
ting processes
dissociation by hard gluons Inverse process leadi
ng to recombination
Notice that in this model dissociation by hard
gluons is considered effective for the J/Y
(contrary to lattice QCD results)
42
Dissociationrecombination at SPS
The smeared form (dashed line) is obtained taking
into account the resolution on NPart, due to our
experimental resolution
Small difference in the ?J/?abs used (4.4 mb)
fixed thermalization time
QGPhadronsregenerationin-medium effects
(Grandchamp, Rapp, Brown)
centrality dependent thermalization time
NA60 In-In _at_ 158 GeV
Pb-Pb _at_ 158 GeV
Qualitatively not too bad
43
19
Dissociationrecombination at RHIC
Only dissociation leads to a suppression stronger
than seen in data
Regeneration can compensate for strong QGP
suppression to come near y0 RHIC data
but this regeneration goes as the (single) charm
density which is poorly known at RHIC
44
Sequential charmonium dissociation (SPS)
Finite temperature QCD predicts that J/Y will
survive up to 1.5 Tc. Namely up to the most
central collisions at SPS energy.
On the other hand, y and cc are dissociated
around Tc. Then we can form the surviving proba
bility
1
The y surviving probability can be determined
from existing data in pA and Pb-Pb (there is not
enough statistics in In-In)
The y suppression pattern in S-U and in Pb-Pb
shows a significantly stronger drop than expected
from the Glauber extrapolation of the p-A data
45
The cc surviving probability in unknown. If we
assume that it is the same as for the y, then we
have
And this seems to agree with the data with a
saturation at 0.6 for the most central collisions
where only the directly produced J/Ys are left.
Caveat will recent HERA-B data on feed-down be
confirmed (?c/J/? 0.2) ? Even more important
if this all true, then at RHIC should wee see
more or less a plateau at energy densities beyond
the highest possible at SPS?
46
20
Sequential melting (SPSRHIC)
RAA/CNM vs. Bjorken energy density
?0 1 fm/c _at_ SPS? 1.6 fm/c crossing time ?0 s
maller _at_ RHIC?
F. Karsch et al., PLB, 637 (2006) 75
47
21
Sequential melting (SPSRHIC)
RAA/CNM vs. Bjorken energy density
?0 1 fm/c _at_ SPS? 1.6 fm/c crossing time ?0 s
maller _at_ RHIC?
Bar uncorrelated error Bracket correlated erro
r Global error 12 is not shown here. Box un
certainty from CNM effects
F. Karsch et al., PLB, 637 (2006) 75
dET/dy PHENIX, PRC 71, 034908 (2005)
48
22
Sequential melting (SPSRHIC)
RAA/CNM vs. Bjorken energy density
?0 1 fm/c _at_ SPS? 1.6 fm/c crossing time ?0 s
maller _at_ RHIC?
What is the conclusion? Data not consistent with
the picture from sequential melting (melt only cc
and y)? However, -Error is large and need bette
r CNM measurements at RHIC. - Need to measure fe
ed-down contribution at RHIC energy.
Bar uncorrelated error Bracket correlated erro
r Global error 12 and 7 are not shown here.
Box uncertainty from CNM effects
49
Sequential Screening Scenario mid vs forward y
QGP suppression of ?C, ? additional forward
suppression from gluon saturation (CGC)?
nucl-ex/0611020
but approx. flat forward/mid above Npart 100
seems inconsistent forward should drop more for
more central collisions as gluon saturation
increases
50
Regeneration Scenario mid vs forward y
both forward mid rapidity suppressed by QGP
i.e. screening or large gluon density
mid-rapidity suppression reduced by strong rege
neration effect but approx. flat forward/mid s
uppression for Npart100 seems inconsistent with
increasing regeneration increasing QGP
suppression for more central collisions
nucl-ex/0611020
51
pT broadening (or non broadening)
Because of collisions suffered by the parton
before the hard process, its transverse momentum
is broadened. This reflects in a pT broadening of
the produced charmonia
Nc number of collisions before hard
process d0 kick in pT received by
the parton
SPS data are consistent with gluon scattering in
the initial state
If a QGP is formed then the sequential melting
will not affect the pT broadening.
The melting implies that less J/Y survives, but
those that survive show the same initial state
effect.
52
At RHIC there is no dependence of . Maybe a
modest rise at forward y
pT
Expected in models with regeneration (Thews)
53
Quarkonium results final summary
Results available with good statistics at the SPS
and now also at RHIC
Reference processes carefully studied in p-A
collisions (E866, HERA-B, NA50) - Still affected
by statistical uncertainties at RHIC (PHENIX)
Anomalous J/? suppression well established by
NA50, NA60 in central In-In and Pb-Pb collisions
and by PHENIX in Cu-Cu and Au-Au
Hadronic suppression mechanisms have difficulties
in explaining the whole J/? suppression
systematics both at SPS (NA50 vs NA60) and at
RHIC (PHENIX)
Interpretation of the results still ongoing the
match is still open between - J/Y dissociation
regeneration - Sequential melting initial state
CGC
Need comprehensive theoretical work that puts
sequential screening, regeneration, gluon
saturation, forward suppression of open charm,
etc. ALL TOGETHER and considers experimental
uncertainties carefully