quarkonium suppression in p-a & a-a collisions …...this talk: revisiting energy loss processes...

Quarkonium suppression in p-A & A-A collisionsfrom parton energy loss in cold QCD matter

Francois Arleo

LLR Palaiseau & LAPTh Annecy

INT Seattle – October 2014

Francois Arleo (LLR & LAPTh) Parton energy loss in pA & AA collisions INT Seattle – Oct 2014 1 / 33

Outline

MotivationsJ/ suppression data in p A collisions

Revisiting energy lossNew scaling properties from medium-induced coherent radiation

PhenomenologyModel for J/ and ⌥ suppression in p A collisionsComparison with data from SPS to LHCExtrapolation to heavy-ion collisions

References

FA, S. Peigne, 1204.4609, 1212.0434, 1407.5054

w/ R. Kolevatov, 1402.1671

w/ R. Kolevatov, M. Rustamova, 1304.0901


Data on J/ suppression in p A collisions

E866ps = 38.7 GeV

0.5

0.6

0.7

0.8

0.9

1

1.1

0 0.2 0.4 0.6 0.8 1xF

_

J/s E866/NuSea

PHENIXps = 200 GeV

0

0.2

0.4

0.6

0.8

1

-3 -2 -1 0 1 2 3y

RdA

u/pp

(y)

PHENIX 3s = 200 GeV

Strong J/ suppression reported at large xF and y

Weaker suppression in the Drell-Yan process


Data on J/ suppression in p A collisions

E866ps = 38.7 GeV

0.5

0.6

0.7

0.8

0.9

1

1.1

0 0.2 0.4 0.6 0.8 1xF

_

J/s E866/NuSea

Drell-Yan E866/NuSea

PHENIXps = 200 GeV

0

0.2

0.4

0.6

0.8

1

-3 -2 -1 0 1 2 3y

RdA

u/pp

(y)

PHENIX 3s = 200 GeV

Strong J/ suppression reported at large xF and y

Weaker suppression in the Drell-Yan process


Interpretations

Many explanations suggested . . . yet none of them fully satisfactory

Nuclear absorptionrequires unrealistically large cross section

nPDF e↵ects and saturationconstrained by Drell-Yan

Intrinsic charmassuming a large amount of charm in the proton


Interpretations





All these e↵ects may lead to some J/ suppression

but cannot alone explain current p A data


Interpretations





All these e↵ects may lead to some J/ suppression

but cannot alone explain current p A data

This talk: revisiting energy loss processes in a simple approach


Gavin–Milana model

Simple model assuming (mean) energy loss scaling like parton energy[ Gavin Milana 1992 ]

�E / E L M

�2

for both Drell-Yan and J/ (though larger due to final-state energy loss)


Gavin–Milana model

Simple model assuming (mean) energy loss scaling like parton energy[ Gavin Milana 1992 ]

�E / E L M

�2

for both Drell-Yan and J/ (though larger due to final-state energy loss)

CaveatsAd hoc assumption regarding E , L, and M dependence of partonenergy loss, no link with induced gluon radiation

Failure to describe ⌥ suppression

�E / E claimed to be incorrect in the high energy limit due touncertainty principle — so-called Brodsky-Hoyer bound


A bound on energy loss ?

Considering an asymptotic charge in a QED model [ Brodsky Hoyer 93 ]

No contribution from large formation times tf � L

Induced gluon radiation needs to resolve the medium

tf ⇠ !

k

2?

. L ! . k

2? L ⇠ q L

2

Bound independent of the parton energyEnergy loss cannot be arbitrarily large in a finite mediumApparently rules out energy loss models as a possible explanation

However

Not true in QED when the charge is deflected

Not necessarily true in QCD due to color rotation


Revisiting energy loss scaling properties

Coherent radiation (interference) in the initial/final state crucial for tf � L

IS and FS radiation cancels out in the induced spectrum

Interference terms do not cancel in the induced spectrum !

Induced gluon spectrum dominated by large formation times

�E =

Zd! !

dI

d!

��ind

= Nc↵s

q�q

2?

M?

E


Two regimes

Incoherent energy loss (small formation time tf ⇠ L)

�E / ↵s q L

2

No color flow in the initial or final state

Large angle particle production

Hadron production in nuclear DIS or Drell-Yan in p A collisions

Coherent energy loss (large formation time tf � L)

�E / ↵s

pq L

M?

E

Needs color in both initial & final state

Important at all energies, especially at large rapidity

Hadron production in p A collisions


Phenomenology

Goal

Explore phenomenological consequences of coherent energy loss

Approach as simple as possible with the least number of assumptions

Observable: J/ and ⌥ suppression in p A collisions

Compare to all available p A datarapidity and transverse momentum dependencepredictions for the p Pb run at the LHC

Provide baseline predictions in heavy-ion collisions


Model for quarkonium suppression

Physical picture and assumptions

Color neutralization happens on long time scales: toctet

� t

hard

Medium rescatterings do not resolve the octet cc pair

Hadronization happens outside of the nucleus: t & L

cc pair produced by gluon fusion


Model for quarkonium suppression

Energy shift

1

A

d� pA

dE

�E ,

ps

�=

Z "max

0d"P(",E )

d� pp

dE

�E + ",

ps

�

Ingredients

pp cross section fitted from experimental data

E

d� pp

dE

=d� ppdy

/✓1� 2M?p

s

cosh y

◆n(ps)

Length L given by Glauber model for minimum bias and centralitydependence

P(✏): probability distribution (quenching weight)


Quenching weight

Usually one assumes independent emission ! Poisson approximation

P(✏) /1X

n=0

1

n!

"nY

i=1

Zd!i

dI (!i )

d!

#�

✏�

nX

i=1

!i

!

However, radiating !i takes time tf (!i ) ⇠ !i��q

2? � L

For !i ⇠ !j ) emissions i and j are not independent


Quenching weight

Usually one assumes independent emission ! Poisson approximation

P(✏) /1X

n=0

1

n!

"nY

i=1

Zd!i

dI (!i )

d!

#�

✏�

nX

i=1

!i

!

However, radiating !i takes time tf (!i ) ⇠ !i��q

2? � L

For !i ⇠ !j ) emissions i and j are not independent

For self-consistency, constrain !1 ⌧ !2 ⌧ . . . ⌧ !n

P(✏) ' dI (✏)

d!exp

⇢�Z 1

✏d!

dI

d!

�!dI

d!

��ind

' Nc↵s

⇡ln

✓1 +

E

2qL

!2M

2?

◆

P(✏) scaling function of ! =pqL/M? ⇥ E


Transport coe�cient

q related to gluon distribution in a proton [ BDMPS 1997 ]

q(x) =4⇡2↵sCR

N

2c � 1

⇢ xG (x , qL)

For simplicity we assume

q(x) = q0

✓10�2

x

◆0.3

(q frozen at x & 10�2)

q0 ⌘ q(x = 10�2) only free parameter of the model

q(x) related to the saturation scale: Q2s (x , L) = q(x)L [ Mueller 1999 ]


Procedure

1 Fit q0 from J/ E866 data in p W collisions

2 Predict J/ and ⌥ suppression for all nuclei and c.m. energies

0

0.2

0.4

0.6

0.8

1

-0.2 0 0.2 0.4 0.6 0.8 1

xF

RW

/Be(x

F)

E866 √s = 38.7 GeV

∧q0 = 0.075 GeV2/fm (fit)

q0 = 0.075 GeV2/fm

Corresponds to Q

2s (x = 10�2) = 0.11 – 0.14 GeV2 consistent with fits

to DIS data [ Albacete et al AAMQS 2011 ]


Procedure

1 Fit q0 from J/ E866 data in p W collisions

2 Predict J/ and ⌥ suppression for all nuclei and c.m. energies

0

0.2

0.4

0.6

0.8

1

-0.2 0 0.2 0.4 0.6 0.8 1

xF

RW

/Be(x

F)

E866 √s = 38.7 GeV

∧q0 = 0.075 GeV2/fm (fit)

0

0.2

0.4

0.6

0.8

1

-0.2 0 0.2 0.4 0.6 0.8 1

xF

RF

e/B

e(x

F)

E866 √s = 38.7 GeV

∧q0 = 0.075 GeV2/fm

Fe/Be ratio well described, supporting the L dependence of the model


SPS predictions

0

0.2

0.4

0.6

0.8

1

1.2

RP

t/p(x

F)

NA3 √s = 19.4 GeV

∧q0 = 0.075 GeV2/fm

0

0.2

0.4

0.6

0.8

1

1.2

RP

t/p(x

F)

NA3 √s = 16.7 GeV (π- beam)

0

0.2

0.4

0.6

0.8

1

0 0.2 0.4 0.6

xF

RP

t/p(x

F)


0

0.2

0.4

0.6

0.8

1

0 0.2 0.4 0.6 0.8

xF

RP

t/p(x

F)


Agreement even at small xFNatural explanation from the di↵erent suppression in p A vs ⇡ A


HERA-B predictions

0

0.2

0.4

0.6

0.8

1

1.2

1.4

-0.4 -0.3 -0.2 -0.1 0 0.1 0.2

xF

RW

/C(x

F)

HERA-B √s = 41.5 GeV

∧q0 = 0.075 GeV2/fm

Also good agreement in the nuclear fragmentation region (xF < 0)

Enhancement predicted at very negative xF


Uncertainties

Two sources of uncertainties are identified

Transport coe�cient q0 (default 0.075 GeV2/fm) to be varied from0.07 to 0.09 GeV2/fm

Parameter (“slope”) of the pp cross section to be varied within itsuncertainty extracted from the fit of pp data


Uncertainties




Uncertainty band determined from the independent variation of q0 and n

(4 error sets)

��R

+�2

=X

k=q0,n

⇥max

�R(S+

k )� R(S0),R(S�k )� R(S0), 0

⇤2

��R

��2 =X

k=q0,n

⇥max

�R(S0)� R(S+

k ),R(S0)� R(S�k ), 0

⇤2


Uncertainties




Largest uncertainty comes from the variation of q0 aroundmid-rapidity

At very large rapidity (e.g. y & 4 at LHC), uncertainty coming from n

becomes comparable or larger than that coming from q0


RHIC predictions

y-3 -2 -1 0 1 2 3

(y)

dAu

R

0

0.2

0.4

0.6

0.8

1

1.2

1.4

E. lossPHENIX

= 0.2 TeV s ψd-Au J/

Good agreement for RpA vs rapidity

Rather small uncertainty coming from the variation of the pp crosssection and the transport coe�cient


p? dependence

Most general case

1

A

d� pA

dE d

2~p?

=

Z

"

Z

'P(",E )

d� pp

dE d

2~p?

(E+",~p? ��~p?)

pp cross section fitted from experimental data

d� pp

dy d

2~p?

/✓

p

20

p

20 + p

2?

◆m

⇥✓1� 2M?p

s

cosh y

◆n

Overall depletion due to parton energy loss

Possible Cronin peak due to momentum broadening

R

pA(y , p?) ' R

loss

pA (y , p?) · Rbroad

pA (p?)


p? dependence at E866

0

0.2

0.4

0.6

0.8

1

1.2

1.4

Rp

Fe

/pB

e(p

T)

Fe / Be ⟨xF⟩ = 0.308

0

0.2

0.4

0.6

0.8

1

1.2

1.4

Rp

W/p

Be(p

T)

W / Be ⟨xF⟩ = 0.308

0

0.2

0.4

0.6

0.8

1

1.2

0 2 4pT (GeV)

Rp

Fe

/pB

e(p

T)

Fe / Be ⟨xF⟩ = 0.48

0

0.2

0.4

0.6

0.8

1

1.2

0 2 4 6pT (GeV)

Rp

W/p

Be(p

T)

W / Be ⟨xF⟩ = 0.48

E866 √s = 38.7 GeV E. loss + broad. Broad. only

Good description of E866 data (except at large p? and large xF)

Broadening e↵ects only not su�cient to reproduce the data


p? dependence at RHIC

0

0.5

1

1.5

2

Rd

Au

/pp(p

T)

Centrality 0-20%0

0.5

1

1.5

2

Rd

Au

/pp(p

T)

Centrality 20-40%

0

0.5

1

1.5

0 2 4 6pT (GeV)

Rd

Au

/pp(p

T)

Centrality 40-60%0

0.5

1

1.5

0 2 4 6 8pT (GeV)

Rd

Au

/pp(p

T)

Centrality 60-88%

y = [ -2.2 ; -1.2 ]

Good description of p? and centrality dependence at y = �1.7


p? dependence at RHIC

0

0.5

1

1.5

2

Rd

Au

/pp(p

T)

Centrality 0-20%0

0.5

1

1.5

2

Rd

Au

/pp(p

T)

Centrality 20-40%

0

0.5

1

1.5

0 2 4 6pT (GeV)

Rd

Au

/pp(p

T)

Centrality 40-60%0

0.5

1

1.5

0 2 4 6 8pT (GeV)

Rd

Au

/pp(p

T)

Centrality 60-88%

y = [ 1.2 ; 2.2 ]

Good description of p? and centrality dependence at y = 1.7


LHC predictions

y-5 -4 -3 -2 -1 0 1 2 3 4 5

(y)

pPb

R

0

0.2

0.4

0.6

0.8

1

1.2

1.4

ψE. loss J/

ΥE. loss

ψE. loss J/

ΥE. loss

= 5 TeV sp-Pb

Moderate e↵ects (⇠ 20%) around mid-rapidity, smaller at y < 0

Large e↵ects above y & 2 – 3

Slightly smaller suppression expected in the ⌥ channel


LHC predictions

y-5 -4 -3 -2 -1 0 1 2 3 4 5

(y)

pPb

R

0

0.2

0.4

0.6

0.8

1

1.2

1.4

E. lossALICE (1308.6726)

= 5 TeV s ψp-Pb J/

Very good agreement despite large uncertainty on normalization

Data at y & 4 would be helpful


LHC predictions

Comparing to other model predictions [ ALICE 1308.6726 ]

Forward J/ suppression underestimated using EPS09 NLO

Forward J/ suppression overestimated in the CGC calculation


LHC predictions

Transverse momentum dependence [ ALICE 1308.6726 ]

R

FB

(p?): good agreement, better agreement with energy losssupplemented by nPDF e↵ects


Extrapolation to heavy-ion collisions

The model successfully reproduces all p A (⇡ A) data vs y and p?

! can be used to predict J/ suppression in heavy-ion collisions

Naturally

Many other e↵ects possibly at work: Debye screening, recombination,energy loss in hot medium. . .

Goal: to set a baseline for the e↵ects of energy loss in cold QCDmatter



Model for A B collisions

Both incoming (projectile & target) partons lose energy in the (target& projectile) nucleus, respectively

Two distinct regions of phase space for gluon emission ! nointerference e↵ects in the radiation induced by nucleus A and B






1

A B

d� ABdy

�y ,

ps

�=

Zd �yB PB("B , y)

Zd�yA PA("A,�y)

d� pp

dy

�y + �yB � �yA,

ps

�

with �yB defined as E (y + �yB) ⌘ E (y) + ✏B






1

A B

d� ABdy

�y ,

ps

�=

Zd �yB PB("B , y)

Zd�yA PA("A,�y)

d� pp

dy

�y + �yB � �yA,

ps

�

A good approximation (at not too large y)

RAB (+y) ' RAp(+y)⇥ RpB (+y) = RpA(�y)⇥ RpB (+y)


Rapidity dependence in A A collisions

y-6 -4 -2 0 2 4 6

pA .

RAp

, RAAR

0

0.2

0.4

0.6

0.8

1

1.2

ψ = 200 GeV J/sAu-Au

ψ = 2.76 TeV J/sPb-Pb

ψ = 200 GeV J/sAu-Au

ψ = 2.76 TeV J/sPb-Pb

y-6 -4 -2 0 2 4 6

pA .

RAp

, RAAR

0

0.2

0.4

0.6

0.8

1

1.2

Υ = 200 GeV sAu-Au

Υ = 2.76 TeV sPb-Pb

Υ = 200 GeV sAu-Au

Υ = 2.76 TeV sPb-Pb

Rather pronounced suppression, especially for J/

RAA slightly decreasing at not too large y

Fast increase at edge of phase space due to energy gain fluctuations


Rapidity dependence in A A collisions at RHIC

y-3 -2 -1 0 1 2 3

CuC

uR

0

0.2

0.4

0.6

0.8

1

1.2

= 200 GeV s 0-20% Cu-Cu ψJ/

E. loss 12 %)±PHENIX (syst global =

= 200 GeV s 0-20% Cu-Cu ψJ/

E. loss 12 %)±PHENIX (syst global =

y-3 -2 -1 0 1 2 3

AuAu

R

0

0.2

0.4

0.6

0.8

1

1.2

= 200 GeV s 0-20% Au-Au ψJ/

E. loss

10 %)±PHENIX (syst global =

Disagreement in both Cu Cu and Au Au collisions

Disagreement more pronounced in Au Au collisions


Centrality dependence in A A collisions at RHIC

partN0 20 40 60 80 100 120

CuC

uR

0

0.2

0.4

0.6

0.8

1

1.2

E. loss y = 0

12 %)±PHENIX |y|<0.35 (syst glob =

= 200 GeV s Cu-Cu ψJ/

partN0 20 40 60 80 100 120

CuC

uR

0

0.2

0.4

0.6

0.8

1

1.2

E. loss y = 1.7

8 %)±PHENIX 1.2 < |y| < 2.2 (syst glob =

= 200 GeV s Cu-Cu ψJ/

Disagreement only in most central Cu Cu collisions


Centrality dependence in A A collisions at RHIC

partN0 50 100 150 200 250 300 350 400

AuAu

R

0

0.2

0.4

0.6

0.8

1

1.2

1.4

E. loss y = 0

12 %)±PHENIX |y| < 0.35 (syst glob =

= 200 GeV s Au-Au ψJ/

partN0 50 100 150 200 250 300 350 400

AuAu

R

0

0.2

0.4

0.6

0.8

1

1.2

1.4

E. loss y = 1.7

9.2 %)±PHENIX 1.2 < |y| < 2.2 (syst glob =

= 200 GeV s Au-Au ψJ/

Disagreement only in most central Cu Cu collisions

Strong disagreement in most central Au Au collisions, fair agreementwithin uncertainties in peripheral collisions


Rapidity dependence in Pb Pb collisions at LHC

y-5 -4 -3 -2 -1 0 1 2 3 4 5

PbPb

R

0

0.2

0.4

0.6

0.8

1

1.2

ψE. loss J/

ΥE. loss

8 %)± (syst global = ψALICE J/

ψE. loss J/

ΥE. loss

8 %)± (syst global = ψALICE J/

= 2.76 TeV s0-90% Pb-Pb

Very good agreement with ALICE data, except in the largest y bins

No hot medium e↵ects ? Or medium e↵ects compensate ?


Centrality dependence in Pb Pb collisions at LHC

partN0 50 100 150 200 250 300 350 400

PbPb

R

0

0.2

0.4

0.6

0.8

1

1.2

E. loss y = 0

13 %)±ALICE |y| < 0.8 (syst global =

= 2.76 TeV s Pb-Pb ψJ/

partN0 50 100 150 200 250 300 350 400

PbPb

R

0

0.2

0.4

0.6

0.8

1

1.2

E. loss y = 3.25

15 %)±ALICE 2.5 < y < 4 (syst global =

= 2.76 TeV s Pb-Pb ψJ/

Excellent agreement with ALICE J/ data



partN0 50 100 150 200 250 300 350 400

PbPb

R

0

0.2

0.4

0.6

0.8

1

1.2

E. loss y = 3.25

E. loss y = 0

13.6 %)±CMS |y|<2.4 (syst global =

= 2.76 TeV s Pb-Pb Υ

E. loss y = 3.25

E. loss y = 0




Disagreement with CMS ⌥ data



partN0 50 100 150 200 250 300 350 400

PbPb

R

0

0.2

0.4

0.6

0.8

1

1.2

E. loss y = 3.25

E. loss y = 0



E. loss y = 3.25

E. loss y = 0




Disagreement with CMS ⌥ data

Indication of hot suppression medium e↵ects for ⌥

. . . implying (?) hot enhancement medium e↵ects for J/


nPDF e↵ects

nPDF e↵ects may a↵ect quarkonium suppression in p A & A Acollisions and could be added (incoherently) to present energy losse↵ects

However sill large uncertainty on small x gluon shadowing (within asingle set or comparing existing sets)

For simplicity we provided “energy loss only” calculations


nPDF e↵ects

Ratio of gluon densities (using EPS09 NLO, x1, x2 given by 2 ! 1 kin.)

y-3 -2 -1 0 1 2 3

(y)

AuAu

R

0

0.2

0.4

0.6

0.8

1

1.2

EPS09DSSZEPS09DSSZ = 200 GeV s ψJ/EPS09DSSZ

y-5 -4 -3 -2 -1 0 1 2 3 4 5

(y)

PbPb

R

0

0.2

0.4

0.6

0.8

1

1.2

EPS09DSSZEPS09DSSZ = 2.76 TeV s ψJ/EPS09DSSZ

At RHIC, energy loss is the leading e↵ect

At LHCEnergy loss leading e↵ect as compared to DSSZSame order of magnitude as EPS09 around mid-rapidity but leadinge↵ect at large rapidity


Summary

Energy loss �E / E due to coherent radiation

Parametric dependence of dI/d! predicted and used forphenomenology

Phenomenology of quarkonium suppression in p A collisions

Good agreement with all existing data vs. y and p?, from SPS to LHCNatural explanation for the large xF J/ suppressionPredictions in good agreement with LHC pPb data

Phenomenology of quarkonium suppression in A A collisions

Model extrapolated from p A to AA collisionsDisagreement observed for J/ at RHIC, especially in most centralcollisions and heavier systemsExcellent (accidental?) agreement observed for J/ at LHC,disagreement observed for ⌥


Medium-induced gluon spectrum

Gluon spectrum dI/d! ⇠ Bethe-Heitler spectrum of massive (color) charge

!dI

d!

��ind

=Nc↵s

⇡

(ln

✓1 +

E

2�q

2?

!2M

2?

◆� ln

1 +

E

2⇤2QCD

!2M

2?

!)

�E =

Zd! !

dI

d!

��ind

= Nc↵s

p�q

2? � ⇤QCD

M?

E

�E / E neither initial nor final state e↵ect nor ‘parton’ energy loss:arises from coherent radiation

Physical origin: broad tf interval : L, thard ⌧ tf ⌧ toctet formedium-induced radiation


Fit to pp data

10-4

10-2

1

0 0.25 0.5 0.75xF

dσ

/dx F

(µb

/A)

E789 p-Be √s = 38.7 GeV

n = 4.5 ± 0.05

0

2

4

-0.4 -0.2 0 0.2xF

dσ

/dx F

(a.u

.)

HERA-B p-C √s = 41.5 GeV

n = 5.7 ± 0.2

0

1

2

-0.2 -0.1 0 0.1 0.2

xF

B d

σ/d

x F (

µb)

PHENIX p-p √s = 200 GeV

n = 8.3 ± 1.1

102

103

104

0 0.01 0.02

xF

dσ

/dx F

(µb

)

ALICE p-p √s = 7 TeV

n = 32.3 ± 7.5


Fit to pp data

10-4

10-3

10-2

10-1

1

10

102

0 10 20 30 40pT (GeV)

Bµµ

dσ

/dp

T (

nb/G

eV

)ATLAS √s = 7 TeV

0.75 ≤ |y| ≤ 1.5

J/ψ

1

10

102

103

104

0 5 10 15pT (GeV)

dσ

/dp

T (

nb/G

eV

)

LHCb √s = 7 TeV

2 ≤ |y| ≤ 2.5

J/ψ

10-4

10-3

10-2

10-1

1

10

0 2 4 6 8pT (GeV)

Bµµ

1/(

2πp

T)

dσ

/dp

T (

nb/G

eV

2)

PHENIX √s = 200 GeV

1.2 ≤ |y| ≤ 2.2

J/ψ10

102

0 5 10 15pT (GeV)

Bµµ

d

σ/d

pT (

pb/G

eV

) LHCb √s = 7 TeV2 ≤ |y| ≤ 2.5

ϒ


quarkonium suppression in p-a & a-a collisions …...this talk: revisiting energy loss processes...

Documents