J/y as a signal of deconfinement
Focus on J/y production results for p+p, d+Au, Au+Au
and Cu+Cu at RHIC (next talk: results with fixed target).
David Silvermyr, ORNL
Critical Point and
Onset of Deconfinement
Firenze, July 4th 2006
Heavy Quarkonia - Intro
Lattice QCD results show that the confining potential between heavy
quarks is screened at high temperature.
Lattice QCD calculation
c
r
c
Color Screening
This screening should suppress bound states such as J/y.
However, recent lattice results indicate that the J/y spectral functions only
show modest modification near the critical temperature, and thus may not
2
be suppressed until higher T.
Original Signature: Matsui & Satz (’86 & ’06)
SPIRES : 934 citations so far (June ’06)
3
An Unambiguous Signature?
• Matsui and Satz carefully
outlined the conditions that
needed to be met for an
observed suppression to be
an unambiguous signature of
QGP formation.
• Focus on one of these
assumptions - may well be
violated. .
4
Competing J/y Production Effects
1.
Normal nuclear absorption:
2.
Shadowing:
3.
Color Screening:
4.
Comover Interactions:
5.
Parton Induced Dissociation:
6.
J/y Recombination:
7.
Feed-down effects, and more..
J/y breakup by nucleons in the final state resulting in charm hadrons
Accounts for parton distribution modifications relative to free protons
Affects parton distribution function before collision occurs
In deconfined medium resonance interactions needed to convert cc pairs
to J/y’s are prohibited
J/y interactions with secondary hadrons results in dissociation
Suppression mechanism that does not require deconfined medium
Breakup of J/y due to in medium parton interactions
Regeneration of J/y’s from off-diagonal c and c pairs
A complex story: the devil is in the details..
5
Observation at CERN SPS (NA50/60)
Pb+Pb collisions show
suppression in excess of
"normal" nuclear suppression
(Recent news: NA60 observed
very similar trend in In+In
collisions.)
Expectation
Suppression
J/y normalized to
Drell-Yan vs “Centrality”
N.B.: D-Y is not the optimal
normalization, closed/open charm
is better.
6
CDF pp (s = 1.8 TeV) results
• Color singlet model
underpredicts high-pT yield.
• Color octet model
overpredicts transverse
polarization at high pT.
T. Affolder et al.,
Phys. Rev. Lett. 85, 2886.
F. Abe et al.,
Phys. Rev. Lett. 79, 572.
7
J/y @ RHIC: Physics Plan
• pp collisions
– Reference, Initial production mechanism
RHIC: can have same √sNN energy as pA and AA..
• pA (or dA) collisions
– Shadowing
– Initial state energy loss
– Cold medium absorption
Many competing effects:
- Reference data essential!
• AA + Light ion collisions
– Modify path length through medium
– Most efficient way to dial in Ncoll,Npart
• Energy scans
– Modify energy density
– More difficult (both luminosity & cross-sections fall quickly
w/ energy)
8
STAR Preliminary
J/y Run4 AuAu
J. Gonzalez, SQM06
Charmonium and Beyond in STAR
STAR Preliminary
J/y Run5 pp
Dielectron Invariant Mass (GeV/c2)
Dielectron Invariant Mass (GeV/c2)
Signal
RHIC Exp.
(Au+Au)
J/y →e+eJ/y →m+m-
PH ENIX
U→ e+e U→ m+m-
STAR
PHENIX
RHIC I
(>2008)
RHIC II
LHC
ALICE+
3,300
29,000
45,000
395,000
9,500
740,000
830
80
11,200
1,040
2,600
8,400
STAR AuAu
preliminary
Nice start with clear mass-peaks for AuAu and pp!
[part of total dataset analyzed]
Rest of talk: focus on PHENIX.
9
Year
RHIC Scaling Law : J/y in PHENIX
Ions
sNN
Luminosity
Detectors
J/
2000
[Run-1]
Au+Au
130 GeV
1 mb-1
Central
(electrons)
~0
2001
Au+Au
200 GeV
24 mb-1
Central
13 + ~0
2002
[Run-2]
p+p
200 GeV
0.15 pb-1
+ 1 muon arm
46 + 66
2002
d+Au
200 GeV
2.74 nb-1
Central
300+800+600
2003
[Run-3]
p+p
200 GeV
0.35 pb-1
2004
[Run-4]
Au+Au
200 GeV
62 GeV
~240 ub-1
~9 ub-1
2005
[Run-5]
Cu+Cu
p+p
200 GeV
200 GeV
~3 nb-1
~3 pb-1
+ 2 muon arms
100+300+120
Central
+ 2 muon arms
~500+2000+2000
Central
+ 2 muon arms
~1000+5000+5000
~1000+5000+5000
Order of magnitude improvements for approx. every two RHIC runs – quite
remarkable (another factor 3 for pp from Run5 to Run6) !
Hope to see continued progress and success like this!
10
Start: p+p Reference
 Consistent with
trend of
world’s data
 ~Consistent
with at least
one COM (Color
Octet Model)
calculation
Phys. Rev. Lett. 96, 012304 (2006).
[Factor x10,
and x30 more
statistics from
Runs5 and 6]
11
d+Au: Disentangle Cold Nuclear Effects
•
Gluon (anti-)shadowing
•
Nuclear absorption.
•
Initial state energy loss.
•
J/y in
South
y<0
X1
X2
X1
X2
rapidity y
Cronin effect
gluons in Pb / gluons in p
J/y in
North
y>0
South (y < -1.2) : via m+m• large X2 (in gold) ~ 0.090
Central (y ~ 0) : via e+e• intermediate X2
~ 0.020
• small X2 (in gold)
~ 0.003
North (y > 1.2) : via m+m-
Shadowing
Anti
Shadowing
X
12
Eskola, et al., Nucl. Phys. A696 (2001) 729-746.
J/y rapidity distribution in p+p and d+Au Collisions
p-p J/Psi – PHENIX 200GeV
R. Vogt: EKS98
shadowing. 3mb
absorption
Rapidity
Total cross section in p+p
(nucl-ex/0507032):
2.61+/-0.20(fit)+/-0.26(abs) µb
X1
J/y in
South
y<0
X2
13
Rapidity and Ncoll Dependence of RdAu: Gluon Shadowing and Nuclear Absorption
RdA   dA /( 2 197   pp )
dA
Yield inv
RdA 
pp
 N coll  Yield inv
1.2
1.0
RdA
0.8
0.6
0.4
0.2
0
Rapidity
• Data favor weak shadowing and weak nuclear absorption effect:
Calc. with 1-3 mb most successful at describing the data.
[Shape reminiscent to what’s seen for dNch/dh (e.g. PHOBOS)]
14
• More suppression for more central events(?)
RUN5 pp News
PHENIX accumulated ~3pb-1 p+p collision during 2005 run. Will give order of magnitude
stat. improvement for reference for d+Au and Au+Au.
Phenix muon arm
Different Quarkonia states test
the degree of color screening
and measure the temperature.
Significant yields (>hundreds)
at RHIC-II ?
1st Upsilons at RHIC !
Beauty measurements will be quite interesting.
15
Heavy Ions: J/y signal in Au+Au
0-20%
20-40%
40-93%
J/ye+e-
Example Mass-plots:
●
●
PHENIX
J/ym+m-
Background subtracted using event mixing
Cu+Cu signal is similar to Au+Au peripheral,
with much larger statistics
16
RAA vs Ncoll (QM’05; nucl-ex/0510051)
About a factor 3 suppression for most central Au+Au points
J/y mm
muon arm
1.2 < |y| < 2.2
J/y ee
Central arm
-0.35 < y < 0.35
dAu
AuAu
CuCu
AuAu
CuCu
mm
mm
mm
ee
ee
200 GeV/c 200 GeV/c 200 GeV/c 200 GeV/c 200 GeV/c
CuCu
mm
62 GeV/c
Band around 1.0
refers to the
uncertainty of the
p+p reference.
[and sometimes has
a global sys. error
added for the
dataset in question..]
17
Results in A+A : vs cold nuclear
matter effects
AA
AA
RAA
dN J/ψ
AB

J/ψ
dN pp
 N coll
suppression
factor ~ 3
suppression
factor ~ 3
1 mb
1 mb
3 mb
3 mb
suppression
factor ~ 2
|y|~1.7
suppression
factor ~ 2
|y|~0
Observe a suppression of ~3 from pp and ( ~ factor 2) beyond cold nuclear
effets. Note common error boxes now (post QM05) around individual points.
18
.. Working on final results with reduced systematic errors ..
Au+Au and Cu+Cu results
On the experimental point of view :
Suppression at RHIC similar to suppression at SPS
Although √s@RHIC=200 GeV and √s@SPS<20 GeV
RHIC
RHICCold
ColdNuc
NucEff
Eff1mb
1mb
Unclear if cold nuclear effects should be :
SPS
= 4.18 mb
SPSabs
abs = 4.18 mb
RHIC Cold Eff 3mb
• different (different suppression pattern)
• or not
(same suppression pattern)
Need more precise measurement of cold
nuclear effect at RHIC
 need more dAu as well as AuAu data
SPS normalized to NA51 p+p value (NA60
preliminary points from Arnaldi, QM05).
19
J/y : Suppression Models
AA
Some suppression
models which
reproduce NA50
data…
… Overestimate the
suppression at PHENIX
Direct suppression in a hot medium :
Cu+Cu
Au+Au
(Hadronic?) co-mover scattering
20
J/ Suppression in Heavy-ion Collisions
May be Masked By Recombination Effects
• In central Au+Au collision there are many (>10) c/cbar pairs produced in a single event.
• Calculations indicate that a significant number of
J/’s could be produced by coalescing c and c-bar
quarks that are the products of different hard
scattering events.
• This would have the effect of masking suppression
due to the presence of a QGP.
21
Comparison with
a prediction w. regeneration
[After update from Rapp et al
to use up-to-date charm and J/y
p+p cross-sections:]
agreement with data points
slightly better than that of
absorption calculation (with 3
mb sigma).
22
• An alternative picture…
23
J/y Feeddown Effect
• J/y yield is populated from both
direct production and feeddown
from the higher resonance states
• Relative yield from each source
experimentally found:
– 60% direct production
– 30% c feeddown
– 10% y’ feeddown
R(  c ) 
N  c * (AJ /y / A c )
N J /y * 
R(  c )  0.32  0.06  0.04

• Medium conditions determine
whether each state exists in the
bound form
Phys.Lett. B561 (2003) 61-72
24
Quarkonia production as a QGP
thermometer
• Even if jet suppression, flow results, etc. have already
established that the medium created at RHIC is an sQGP, we
would still like to establish its properties.
• The quarkonium suppression pattern may be able to serve as a
QGP thermometer.
• In cartoon form…
H. Satz, J. Phys. G32,
R25 (2006)
• It is argued that the common pattern seen at SPS and RHIC is
due to complete suppression of ’ and c, which feeds down to
create ~40% of the J/’s, and that the primordial J/’s aren’t
suppressed at all by screening.
25
CuCu: More Bins...
Copper-Copper 200 GeV
J/y |y| = 1.2-2.2
• Rather smooth onset/scaling with centrality.. no distinct onset or
plateau for c suppression, with our preliminary data & errors
26
Test of Npart scaling
Alternative looks at data may help to break gridlock..
Can the results be explained by some other scenario? Geometry and
surface effects or scaling a la soft processes?
[argued for NA50 data by e.g. Gazdzicki, Braun-Munzinger et al.]
27
More variables : Rapidity
• Rapidity distribution of recombined J/y is supposed to be peaked at
y=0 (e.g. R.L. Thews & al., nucl-th/0505055)
– True IF charm distribution ~ J/y in p+p !
p+p data
pQCD, adjust <kT2>
off-diagonal
(with recomb.)
diagonal
– But Au+Au charm rapidity distributions might be rather flat!
28
Invariant yield vs pt
Cu+Cu (|y|[1.2,2.2])
Au+Au (|y|[1.2,2.2])
2 -6
We fit the pt spectrum using A[1 + ( pt / B) ] to extract <pt2>
29
Mean transverse momentum vs Ncoll
Recombination (Thews
et al., nucl-th/0505055)
predicts a narrower pT
distribution, leading to
a lower <pT²>
Experimentally
:
data
falls
between the two hypotheses.
Need to consider all datasets and
error bars before drawing conclusions.
Open markers : |y|<0.35
Solid markers : |y|~1.7
p+p
d+Au
Open markers : |y|<0.35
Solid markers : |y|~1.7
Cu+Cu
p+p d+Au
Au+Au
With recombination
With recombination
Cronin / Broadening:
pT
2
AA
 pT
2
pp
+Δ p
2
T
ρσ L AA
30
J/y Status
RHIC data exhibits a factor 3 suppression for most central events in Au+Au
collisions. Suppression vs Npart rather similar to what was seen at SPS.
Comparison with models (here only used a subset..) suggests that
1) Models with only cold nuclear matter effects tend to under-predict the suppression
2) Models with color screening or comovers and without recombination have
too much suppression
3) Models with recombination are in rather reasonable agreement with the data
Not clear if recombination is the explanation though. Feed-downs suppressed?
Mixed evidence for recombination from other variables:
Con(?): The rapidity dependence of the J/y yield shows no dramatic change
in shape with increasing Npart.
Pro(?): <pT2> is also consistent with flat behaviour, but large error bars.
31
J/y Action Items
●
Need more work on data (in progress); reduce size of errors and go to final
results. Using the statistically superior Run5 p+p dataset for reference should
be helpful. It would be nice to confront theory with more precise results! :=)
• Flow? - J/y v2 studies started; no results yet. Statistically very challenging
analysis with existing RHIC datasets. Comparison between charm and
charmonium should be instructive.
• Question: Do we see (suppression + recombination) or just not so much
suppression to start with..?
[‘soft’ scaling and similarity with NA50 suppression pattern - somewhat
surprising and hard to overlook. Just coincidences?]
32
More data needed!
• In any case (and as usual..), more data is needed..
• Need to study
– Different quarkonia states (different melting points,
different feeddown contributions).
– Different collision energies
• Modify charm quark density to change recombination fraction.
• Modify temperature.
– Better data vs. centrality, pT, y.
– Polarization, J/-hadron correlations, flow (for production
mechanism).
• This physics is really just getting started at RHIC..
33
Future Measurements: y’
Run 6 200GeV p+p
With more luminosity
we should be able to
measure y’ in AuAu
too!
Invariant Mass
(GeV/c2)
34
Future Measurements: c
PHENIX
Run 5 200GeV p+p
(c - J/y) Mass (GeV/c2)
PHENIX
Run 5 200GeV p+p
(c - J/y) Mass (GeV/c2)
Run 6 data set has a factor of x3 more luminosity.
A very tough measurement in AuAu; dAu probably doable.
35
Exotica: More to Come
Ultra-peripheral Collisions (UPC’s)
UPC’s : well calibrated
EM probe
 measured by STAR
QM05
J/y by PHENIX
36
Future
 Hopefully (PAC willing)….
Run 7 & 8: high statistics Au+Au 200GeV, x10 luminosity
high statistics d+Au 200GeV, x10 luminosity
And comparisons with STAR results!
Longer term:
RHIC Upgrades:
Increased luminosity
Increased species
Detector Upgrades :
Reaction Plane Detector
(PHENIX, from Run-7)
Si Vertex Detector
(PHENIX and STAR)
Nosecone Calorimeter, muon
trigger upgrade, …
Then there are also the LHC experiments soon, and the nice results from NA60 (next),
37
so the upcoming few years should be really interesting!
Near-Term Future..
Let’s hope for some nice and friendly semi-final matches
today and tomorrow (9 PM) !
38
Backup slides
39
Heavy quarkonium states, energy levels and radii
Quarkonium – bound q/q-bar state
40
J/y transport model
Adding QGP hydro and J/y
transport  better agreement
Model includes :
•Detailed QGP hydro
•J/ψ transport
•normal nuclear absorption:
•σabs = 1 mb
•σabs = 3 mb
(Curves for y=2 and y=0 are similar)
Au+Au y~1.7 |y| ~ 0
Zhu, Zhuang, Xu, PLB607 (2005) 107
41
+ private communications
 Based on recent lattice QCD calculations,
J/y melting temperature could be higher than
initially expected  suppression of direct J/y
could be out of the range of RHIC
 On the other hand c and y’ should melt at
a temperature close to TC (~1.1 – 1.2 TC)
 Anomalous suppression comes from c and
y’ feed-down.
Karsch, Kharzeev and Satz, hep-ph/0512239
H. Satz, Hep-ph/0512217
Sequential charmonium dissociation
Quarkonium dissociation temperatures – Digal, Karsch, Satz
Overall J/y survival probability = measured/expected
direct J/y survival probability
assume to be 1 at SPS energy
Feed-down
J/y S
survival
probablity
0.4 S x  0.3
χ + 0.1 S ψ'
c
J/y feed-down :
• ~60% from direct
production
• ~30% c  J/y + 
• ~10% y’  J/y + X
42
Sequential charmonium dissociation
Karsch, Kharzeev and Satz, hep-ph/0512239
SPS data
At SPS, NA50 measured :
• J/y suppression
• y’ suppression
• but not c
NA60 preliminary
 0.6 + 0.4 S ψ'
(Lattice QCD  Sy’~Sc)
Karsch, Kharzeev and Satz, hep-ph/0512239
SPS + RHIC
data
0.6
NA60 preliminary
PHENIX preliminary
At RHIC, PHENIX measured :
• J/y suppression.
data are consistent with sequential
charmonium dissociation at both
RHIC and SPS.
Note: Systematic errors ignored..
More data needed
43
Suppression Mechanism
•
J/y Suppression Models:
– assume heavy quarkonia are formed only
during the initial hard nucleon-nucleon
collisions
– Subsequent interactions only result in
additional loss of yield
•
Color Screening:
– Color charge of one quark masked by the
surrounding quarks
– Prevents cc-bar binding in the interaction
region
– Characterized by Debye screening radius
(rD)
– If the screening radius is smaller than the
J/y radius then the quarks are effectively
masked from one another
c
c
Color Screening
44
RAA vs Npart : Comparison with
NA50 data (QM’05)
NA50 data is normalized to NA50 p+p
point.
Suppression level is rather similar
between the two experiments,
although the collision energy is 10+
times higher at RHIC (200 GeV vs 17
GeV).
Note: size of error bars, or common
systematic error band not negligible!
45
RAA vs Npart: Comparison with
cold nuclear effects (QM05)
Forward rapidity
Mid rapidity
Prediction from pQCD calculations, including 3mb nuclear
absorption and shadowing.
Seems to underestimate the suppression somewhat.
Note: abs somewhat too high wrt d+Au data; Should have
1 mb curve also.
46
RAA vs Npart: Comparison with
predictions without regeneration
(QM05)
Models which approx. reproduce
NA50 data, with J/y suppression
only. (no regeneration mechanism)
Over-estimates J/y
suppression at RHIC!
47
RAA vs Npart : Comparison with
predictions w. regeneration (QM05)
Models using suppression + various
regeneration mechanisms;
Better matching with data points,
but note that all model
calculations should be checked to
use up-to-date charm and J/y
p+p cross-sections!
(reduced exp. errors on those
quantities would also help)
48
The PHENIX detector
Central arms:
hadrons, photons, electrons
J/e+ep > 0.2 GeV/c
|y| < 0.35
  
Muon arms:
muons at forward rapidity
J/m+mp > 2GeV/c
1.2 < |y| < 2.4
  
Centrality measurement:
We use beam beam counters together with zero degree calorimeters
Centrality is mapped to Npart (Ncol) using Glauber model
49
PHENIX Detector: Muon Arms
• Muon Tracker and Muon
Identifier provide good
momentum resolution and
tracking ability
• High rate level 1 dimuon
trigger
• Online level 2 filtering
PHENIX
p+p 200GeV
Like Sign
Subtraction
PHENIX
p+p 200GeV
50
PHENIX Detector: Central Arm
• Drift Chamber provides high
resolution tracking and
momentum resolution
• RICH and EmCal provide
electron identification
• High rate level 1 electron
trigger
• Online level 2 filtering
Like Sign
Subtraction
51
Silicon Vertex Detector
• Four barrel layers
– Two ALICE pixel bus layers
– Two strip-pixel layers
• Four end-cap pixel layers
• Displaced vertex (σ ~50 mm)
• Full azimuthal inner tracking
|η| < ~2.4
– Improve acceptance for -jet
correlations, D  K
• Connect to tracks in central
and muon arms
– Tag heavy flavor decays
• c,b  e,m
• B  J/
– Improve onium resolution
– Eliminate decay hadrons
– Reduce high-pT background
52
Nose Cone Calorimeter
• Replace central arm magnet
nosecones (Cu) w/ tungstensilicon calorimeters
• Coverage at forward/backward
rapidity: 0.9 < |η| < 3.5
– /0 separation for pT < 30 GeV/c
– Jet identification
•  identification gives good
acceptance for c  J/ + 
53
Muon Trigger Upgrade
• Three layers of RPCs with 2D
(θ,φ) pad readout
• Provides online momentum
measurement to improve Level-1
trigger rejection
– Single-particle
• pT cut
• W spin-measurements in pp
– Two-particle
• Minv cut
• onium measurements in AA
– Necessary to take complete
advantage of luminosity upgrades
• Provides improved highmultiplicity background rejection
54
Intermission: Comparison with
Other Prompt Probes
A general way to classify QCD probes is by speed and color multiplet; different
combinations give rise to different classes of high-Q2 observables:
q: fast color triplet
Induced
gluon
radiation?
g: fast color octet
Q: slow color triplet
QQbar: slow color
singlet/octet
Energy
Loss?
Dissociation?
Virtual photon: colorless
Real photon: colorless
(P. Stankus)
Controls
Unknown Medium
55
Why is it hard to draw conclusions from the observed J/y RAA ?
• Compare and contrast J/ vs. jets
– Initial jet production well understood (pQCD vs
data)
– Cold nuclear matter effects for jets give RpA > 1
(opposite of signal, easy to disentangle; also have
direct photons as add’l control.)
Not true for J/’s unfortunately..
56
Jets: AuAu vs. dAu (PHENIX)
Phys. Rev. Lett. 91, 072303 (2003).
Au + Au Experiment
Final Data
d + Au Control Experiment
Preliminary Data
57
Comparison of leading 0 spectrum
(PHENIX) to pQCD
Phys. Rev. Lett. 91, 241803 (2003).
58
Jets Strongly Suppressed at RHIC!
•
Photons are not suppressed  initial state production *does*
scale with Nbinary.
[ From magnitude of jet suppression we are able to quantify the
gluon density and this is viewed as one of the cornerstones of
the arguments that we have created an sQGP at RHIC. ]
59
• Other aspects of “rich” J/ physics…
– Thermal charm enhancement – not
– Charm quark energy loss
– Recombination
60
Total Charm Production Scales w/Nbinary
PPG035
Phys. Rev. Lett. 94, 082301 (2005).
• It had been suggested
that, in addition to being
produced in initial hardscattering events, charm
quarks in RHIC collisions
could also be produced
via collisions of thermal
partons due to the
extreme temperatures
that would be reached.
• PHENIX data shows that
this is apparently not
significant.
61
But, Charm Quarks Lose Energy in the
Medium Created at RHIC
PPG056
Phys. Rev. Lett. 96,
032301 (2006).
62