Heavy quarkonia and Quark-Gluon Plasma: what did we learn and what are we learning?

E. Scomparin (INFN Torino)

Junior Day, ALICE week March 24 – 28, 2014

Charmonium (bottomonium) an observable connected with deconfinement a thermometer of the system

Studied since SPS times (1986 onwards) Lots of results from lower-energy facilities Role of LHC

Outline

Introduction, some concepts, a bit of history Measuring charmonium at ALICE Pb-Pb/Au-Au collisions

SPS and the “anomalous suppression” RHIC, between suppression and regeneration LHC

Charmonia, a decisive test of regeneration scenarii Bottomonia, a decisive test of suppression scenarii

What about cold nuclear matter ? Conclusions

Heavy quarkonium states

Almost 40 years of physics!

Spectroscopy Decay Production In media

See 182 pages review on arXiv:1010.5827

Bound states cc / bb

q

q

Colour screening

4

At T=0, the binding of the 𝑞 and 𝑞 quarks can be expressed using the Cornell potential:

krr

rV

)(

Coulombian contribution, induced by gluonic exchange between 𝑞 and 𝑞

Confinement term

q

q

The QGP consists of deconfined colour charges the binding of a 𝑞𝑞 pair is subject to the

effects of colour screening

What happens to a 𝑞𝑞 pair placed in the QGP?

krr

rV

)( Dre

rrV

/)(

• The “confinement” contribution disappears

• The high color density induces a screening of the coulombian term of the potential

q

q

..and QGP temperature

• Screening stronger at high T

• D maximum size of a bound state, decreases when T increases

Resonance melting

QGP thermometer

• Different states, different sizes

However, charmonium production can proceed, at LHC energies

• directly in the interaction of the initial partons • via the decay of heavier charmonia (feed-down)

• via the decay of B-mesons (non-prompt) ~ 10 % at pT=1.5 GeV/c

~ 8% from (2S) ~25% from c

Feed-down and suppression pattern

J/

(3S) b(2P)

(2S)

b(1P)

(1S)

(2S) c(1P)

J/

Digal et al., Phys.Rev. D64(2001)

094015

• Since each resonance should have a typical dissociation temperature, one should observe «steps» in the suppression pattern of the measured J/ (or (1S)) when increasing T

• Ideally, one could vary T • by studying the same system (e.g. Pb-Pb) at various s • by studying the same system for various centrality classes

Yie

ld(T

)/Yie

ld(T

=0)

Can we get Tdiss ?

Lattice QCD calculations are our main source of information on the dissociation temperatures Early studies showed that the complete disappearance of the J/ peak occurred at very high temperatures (~2Tc) However spectral functions expected to change rather smoothly

How to pin down Tdiss ?

From O. Kaczmarek

Current knowledge of Tdiss

• Binding energies for the various states can be obtained from potential models too • Assume a state “melts” when Ebind < T Tdiss ~1.2 Tc

Ebin Tweak binding

Ebin Tstrong binding

•Recent results from lattice on J/: •No clear sign of bound state beyond T=1.46 Tc

•Region close to Tc now under study

O.Kaczmarek@HP12

From suppression to (re)generation At sufficiently high energy, the cc pair multiplicity becomes large

Contrary to the suppression scenarii described before, these approaches may lead to a J/ enhancement

Statistical approach: Charmonium fully melted in QGP Charmonium produced, together with all other hadrons, at chemical freeze-out, according to statistical weights

Kinetic recombination: Continuous dissociation/regeneration over QGP lifetime

..and the use of charmonia as QGP thermometer becomes questionable

How do we measure quarkonia ?

In the simplest way: through dilepton decay (e+e-, +-) Dimuon decay is the most commonly measured channel (but ALICE has capability for both channels)

Some crucial aspects to take into account when designing an experiment measuring muon pairs

Mass resolution Needed to separate resonances with similar masses

Accurate vertex information Useful to separate prompt from displaced production (and also to reduce various backgrounds)

Possibility to work at high luminosity Necessary in particular for “rare” processes (bottomonium)

Wide acceptance Lots of physics can be learnt with differential studies on large rapidity/transverse momentum windows

Measurement via decays

11

beam

Muon Other

hadron absorber

and tracking

target

muon trigger

magnetic

field

Iron w

all

Place a huge hadron absorber to reject hadronic background

Implement a trigger system, based on fast detectors, to select

Reconstruct tracks in a spectrometer

Extrapolate muon tracks back to the target Vertex reconstruction not optimal (z ~ cm) use “external” vertex constraint if possible

Correct for multiple scattering and energy loss

Approach adopted by NA50, PHENIX and ALICE muon

Measurement via decays

12

Approach adopted by NA60, CMS and foreseen in PHENIX and ALICE upgrades

Dipole magnet

target

vertex tracker

or

!

hadron absorber Muon Other

and tracking muon trigger m

agnetic

field

Iron w

all

Use a tracker (usually Si detectors) in the vertex region to track muons before they suffer multiple scattering and energy loss in the hadron absorber.

Improve mass resolution

Determine origin of the muons

Miglioramento di m

’

NA60

before

after

In-In @ 158 GeV A

NA60

f

p-nucleus at 400 GeV

• From 70 to 20 MeV at mµµ~ 1 GeV

• From 100 to 70 MeV at mµµ~ 3 GeV

Improving the mass resolution

NA60 data (SPS)

Measuring J/ in ALICE (muons)

Forward rapidity measurement in the muon spectrometer

Hadron absorber

5 tracking stations

Iron wall Dipole 2 trigger stations

3 Tm dipole magnet 10 I hadron absorber, 7 I iron wall

Triggering on muons

Track fires MT1 (MT2) in Y1 (Y2) Y2=Y2-Y2, deviation with respect to an infinite momentum track Y2 cut ( Yf) is equivalent to a pT cut (at first order) Y2 stored in look-up tables. Minimum pT cut ~0.5 GeV/c

Reconstructing muon tracks

Build combinations of clusters between chambers in station 5 (4), estimating momenta (hypothesis: track comes from IP) Extrapolate to station 4 (5) and recompute track parameters Remove track candidates sharing the same clusters or those with bending momentum/non-bending impact parameter outside (tuned) reconstruction parameters Propagate to station 3, 2, 1 looking for compatible cluster and recomputing at each step Remove tracks sharing the same clusters on the basis of largest number of clusters or lowest 2 (if equal) Correct for energy loss and multiple scattering in the front absorber and extrapolate to the vertex

Cleaning the event sample Several cuts used to go towards the event sample for physics analysis

Remove tracks with Rabs<17.6 cm or Rabs> 80 cm (where corrections are more delicate)

Remove tracks NOT matching a trigger tracklet (very powerful to remove remaining contamination from hadrons and beam/gas)

Add a DCA-related cut (pDCA, Gaussian distribution for good tracks) to remove fake tracks

Invariant mass spectrum

)2(

))((//

pppppp

pppppp JJ

)cos(22 22

/ ppEEmmJ

Writing the 4-momentum of the muons one easily gets

Let’s consider the 2-body decay J/+-

(it is necessary to measure momentum and opening angle of leptons)

2sin2/

EEmJAt high energy E~p, m ~0

J/ and signals clearly visible after cleaning the event sample

Fitting the invariant mass spectrum

Continuum under the resonances

Superposition of various physics processes

In the charmonium region, mainly Combinatorial background (uncorrelated / K) Open charm/beauty decays

Two possibilities Subtract combinatorial background using like-sign and build hadronic cocktail for other sources or Use empirical shapes for the continuum

NNN 2

We have electrons too…

Completely different analysis approach with respect to muon analysis

PID using TPC information

Inclusion cut for electrons Exclusion cuts for pions and protons

In spite of the various cuts background levels are typically larger than in the muon analysis

Require hit in the innermost pixel layer help in rejecting photon conversions

Decay electrons have pT~1.5 GeV/c in the J/ rest frame require pT>0.85 GeV/c

Background subtraction

In this situation, the so-called “mixed event” approach proves to be more powerful

Details may vary, but in general

1) Bin the events according to their characteristics (kinematic, vertex position, centrality) 2) Build pools of /e of different charge 3) Combine /e from pools to build “mixed event” opposite sign di-muon/-electron spectrum 4) Normalize to measured opposite sign spectrum (various prescriptions possible)

“Classical” approach via like-sign

is often unsatisfactory when S/B is low or statistics are not huge

NNN 2

Signal visible also in central Pb-Pb!

And now… to the results

Although there are a number of significant results at LHC energies, a comprehensive look at past/present results is necessary in order to reach a satisfactory understanding

For a better understanding remember that results are usually given in terms of

Ratio of the charmonia yield to the Drell-Yan dileptons (SPS energy) Nuclear modification factor (RHIC+LHC energy)

q

q

* +

-

Cold nuclear matter effects play an important role! Parameterized from p-A data, but NOT an easy task!

Just a bit of history The prediction of J/ suppression in 1986 led immediately to a strong experimental effort to detect such an effect First results on the same year from O+U collisions by the NA38 experiment at the CERN SPS…

periferiche centrali NJ/

cont. (2.7-3.5)

... suggested a suppression of J/ by a factor~2 moving from peripheral towards central events

Peripheral events

Central events

Do we see a QGP effect in O-U collisions at SPS? Is this the end of the story ?

PbPb results at sNN =17.2 GeV (SPS)

NA50 and the discovery of the anomalous J/ suppression

N.B.: cold nuclear matter effects were calibrated here from pA results obtained at higher s (27.4 GeV)

SPS “summary” plot

After correction for EKS98 shadowing

In-In 158 GeV (NA60) Pb-Pb 158 GeV (NA50)

Compare NA50 (Pb-Pb) and NA60 (In-In) results, after correcting for CNM effects evaluated at the correct s

Anomalous suppression for central PbPb collisions (up to ~30%, compatible with (2S) and c melting) Are we seeing a hint of the sequential suppression originally predicted in the ‘80s ?

B. Alessandro et al., EPJC39 (2005) 335

R. Arnaldi et al., Nucl. Phys. A (2009) 345

QGP-induced suppression or hot hadron gas effect ? Qualitatively, sharp onset at large Npart may favour QGP scenario (hadron gas effects should lead to a smoother Npart dependence)

Is (2S) suppressed too ?

Yes, but already for light-nuclei projectiles (S-U collisions)

Makes sense, the less bound (2S) state may need lower temperatures to melt

Up to now, the most accurate set of results on (2S) production in nuclear collisions

Does this result confirm the sequential suppression scenario ? Not clear, (2S) expected to be completely suppressed…. …which is not the case

Moving to RHIC: expectations

Two main lines of thought

1) We gain one order of magnitude in s. In the “sequential suppression” scenario we have then two possibilities a) We reach T>Tdiss

J/ suppression becomes stronger than at SPS

b) We do not reach T>TdissJ/ suppression remains the same

2) Moving to higher energy, the cc pair multiplicity increases

A (re)combination of cc pairs to produce quarkonia may take place at the hadronization J/ enhancement ?!

J/ RAA: SPS vs RHIC Let’s simply compare RAA (i.e. no cold nuclear effects taken into account)

Qualitatively, very similar behaviour at SPS and RHIC !

PHENIX experiment measured RAA at both central and forward rapidity: what can we learn ?

Do we see (as at SPS) suppression of (2S) and c ?

Or does (re)generation counterbalance a larger suppression at RHIC ?

RHIC: forward vs central y

29

Comparison of results obtained at different rapidities

Stronger suppression at forward rapidities

Mid-rapidity

Forward-rapidity

Not expected if suppression increases with energy density (which should be larger at central rapidity)

Are we seeing a hint of (re)generation, since there are more pairs at y=0?

Or this is just related to CNM effects, different in the two regions ?

Suppression vs recombination Do we have other hints telling us that recombination can play a role at RHIC ?

Recombination could be measured in an indirect way

J/ y distribution should be narrower wrt pp

J/ pT distribution should be softer (<pT

2>) wrt pp

J/ elliptic flow J/ should inherit the heavy quark flow

charm Open

Closed

Difficult to conclude

<pT2> vs system size

No clear decrease of <pT

2> wrt pp at RHIC, as expected in case of recombination

…still, at the SPS, there was a very clear increase from elementary to nucleus-nucleus collisions

Difficult to conclude

Comparisons with models

In the end, models can catch the main features of J/ suppression at RHIC, but no quantitative understanding

In particular, no clear conclusion on

(2S) and c only suppression

vs

All charmonia suppressed + (re)generation

An interesting comparison

What happens if we try taking into account cold nuclear matter effects and compare with the same quantity at the SPS ?

Nice “universal” behavior

Note that 1) charged multiplicity is proportional to the energy density in the collision 2) Maximum suppression ~40-50% (still marginally compatible with only (2S) and c melting)

Is this picture confirmed by LHC data?

In spite of the “difficulties” in understanding CNM effects and extrapolating them to AA…

Great expectations for LHC

34

…along two main lines

1) Evidence for charmonia (re)combination: now or never!

Yes, we can!

(3S) b(2P)

(2S) b(1P)

(1S)

2) A detailed study of bottomonium suppression

No recombination expected here!

Mass r0

J/, ALICE vs PHENIX

35

Compare with PHENIX Stronger centrality dependence at lower energy Systematically larger RAA values for central events in ALICE

Is this the expected signature for (re)combination ?

Even at the LHC, NO rise of J/ yield for central events, but…

RAA vs Npart in pT bins

In the models, ~50% of low-pT J/ are produced via (re)combination, while at high pT the contribution is negligible fair agreement from Npart~100 onwards

J/ production via (re)combination should be more important at low transverse momentum

Different suppression pattern for low- and high-pT J/

Smaller RAA for high pT J/

Compare RAA vs Npart for

low-pT (0<pT<2 GeV/c) and high-pT (5<pT<8 GeV/c) J/

recombination

recombination

36

RAA vs pT

37

Expect smaller suppression for low-pT J/ observed!

The trend is different wrt the one observed at lower energies, where an increase of the <pT> with centrality was obtained

Fair agreement with transport models and statistical model

CMS, focus on high pT Muons need to overcome the magnetic field and energy loss in the absorber Minimum total momentum p~3-5 GeV/c to reach the muon stations Limits J/ acceptance

Midrapidity: pT>6.5 GeV/c Forward rapidity: pT>3 GeV/c

..but not the one (pT > 0 everywhere)

CMS vs STAR high-pT suppression

39

(Re)combination effects should be negligible

CMS: prompt J/ pT > 6.5 GeV/c , |y|<2.4 0-5% factor 5 suppression 60-100% factor 1.4 suppression

STAR: inclusive J/ pT>5 GeV/c , |y|<1

High pT: less suppression at RHIC than at LHC

(exactly opposite behaviour at low pT !)

Does the J/ finally flow ?

40

First hint for J/ flow in heavy-ion collisions (ALICE, forward y) !

The contribution of J/ from (re)combination should lead to a significant elliptic flow signal at LHC energy

Significance up to 3.5 for chosen kinematic/centrality selections

Qualitative agreement with transport models including regeneration

J/ at the LHC: a “summary” plot

41

Main “qualitative” features now explored

Effect of “inclusive” (ALICE) vs “prompt” (CMS) expected to be small

Precise theory calculations are now needed!

“Onset” of regeneration at small y, pT ?

J/ and open charm, more questions

42

Is the apparent similarity of D and J/ RAA telling us something ? In principle suppression mechanisms are different (en. loss vs suppression) but….

(2S): CMS vs ALICE

43

(2S) much less bound than J/ Results from the SPS showed a larger suppression than for J/ (saturating towards central events ? One of the landmarks of stat. model) No results from RHIC in Au-Au Seen by both CMS (much better resolution!) and ALICE, different kinem.

Expectations for LHC ?

Enhancement/suppression ?

44

At SPS, the suppression increased with centrality (the opposite for CMS) Overall interpretation is challenging ALICE vs CMS: should we worry? Probably not, seen the size of the errors Large uncertainties: signal extraction, pp reference Work needed to reduce systematics/ use pp 2.76 data collected in 2013

CMS: transition from strong (relative) enhancement to suppression in a relatively narrow pT range ALICE excludes a large

enhancement

Finally, the

45

LHC is really the machine for studying bottomonium in AA collisions (and CMS the best suited experiment to do that!)

Strong suppression of (2S), wrt to (1S)

Separated ϒ(2S) and ϒ(3S) Measured ϒ(2S)/ϒ(1S)double

ratio vs. centrality centrality integrated

no strong centrality dependence

Upper limit on ϒ(3S) centrality integrated

One of the long-awaited signatures ?

First accurate determination of suppression

47

Suppression increases with centrality First determination of (2S) RAA: already suppressed in peripheral collisions

(1S) compatible with only feed-down suppression ?

Probably yes, also taking into account the normalization uncertainty

Compatible with STAR (but large uncorrelated errors): expected ? Is (1S) dissoc. threshold still beyond LHC reach ? Full energy

Cold nuclear matter

centrality

J//N

coll pA

J//N

coll pA

AA

Various cold nuclear matter effects can affect charmonia Shadowing Initial/final state parton energy loss Pair break-up (negligible at LHC energies)

Studied via p-A collisions Investigate QCD-related effects Extrapolate their contribution to A-A collisions

Some recent ALICE pA results

Backward y

Central y

Forward y

A significant suppression of the J/ is visible at forward and central y

Theoretical models including Shadowing Coherent energy loss

reproduce the main features but not the details of the pT dependence Color Glass Condensate approach clearly overestimates the suppression at forward rapidity

Implications for Pb-Pb results

If shadowing is the main cold nuclear matter effect at play, a simple estimate of CNM influence on Pb-Pb results can be obtained by forming the products

backwardRforwardR pPbpPb centralRpPb

2

Clear “hot matter” suppression at large pT At low pT indications for re-generation (especially at mid-rapidity)

Conclusions

Heavy quarkonia and QGP: after >25 years still a very lively field of investigation, with surprises still possible

Very strong sensitivity of quarkonium states to the medium created in heavy-ion collisions: interpretation not always easy

Two main mechanisms at play in AA collisions 1) Suppression by color screening/partonic dissociation 2) Re-generation (for charmonium only!) at high s

can qualitatively explain the main features of the results

Future of quarkonia multi-differential suppression studies full LHC energy complete characterization of excited states in the QGP

Backup

Size: 16 x 26 meters

Weight: 10,000 tons

Detectors: 18

Focus on Pb-Pb

•Centrality estimate: standard approach

PRL106 (2011) 032301

•Glauber model fits •Define classes corresponding to fractions of the inelastic Pb-Pb cross section

Charged multiplicity – Energy density

• dNch/d = 1584 76

• (dNch/d)/(Npart/2) = 8.3 0.4

• ≈ 2.1 x central AuAu at √sNN=0.2 TeV

• ≈ 1.9 x pp (NSD) at √s=2.36 TeV

• Stronger rise with √s in AA w.r.t. pp

• Stronger rise with √s in AA w.r.t. log extrapolation from lower energies

55

PRL105 (2010) 252301

• Very similar centrality dependence at LHC & RHIC, after scaling RHIC results (x 2.1) to the multiplicity of central collisions at the LHC

PRL106 (2011) 032301

c)GeV/(fm15 2Bj

System size

56

• Spatial extent of the particle emitting source extracted from interferometry of identical bosons

• Two-particle momentum correlations in 3 orthogonal directions -> HBT radii (Rlong, Rside, Rout)

• Size: twice w.r.t. RHIC

• Lifetime: 40% higher w.r.t. RHIC

ALICE: PLB696 (2011) 328 ALICE: PLB696 (2011) 328

Very first result A couple of weeks after the end of the 2010 data taking, ATLAS published the first LHC result on J/ suppression!

Interesting, but a bit deceiving! ~ same suppression as at RHIC,SPS However this comparison is not sound. Different pT explored!

First results at the LHC

Great expectation (as for many other observables) from the first LHC heavy-ion runs (Pb-Pb @ 2.76 TeV)

Advent of upsilon family (seen also at RHIC with small statistics) (Re)generation negligible Observe suppression of less bound (2S), (3S) wrt (1S)

Solve/clarify issue with J/ suppression/(re)generation

Complementary acceptance ALICE low pT J/ CMS/ATLAS high pT J/, excellent resolution

Mass r0

It’s a long story….

60

…27 years after the prediction of J/ suppression by color screening

… 13 years after the prediction of charmonium regeneration

It’s a long story….

61

…27 years after O beams were first accelerated in the SPS

…13 years after Au beams were first accelerated at RHIC

… and barely 2.5 years (!!!) after Pb beams first circulated inside the LHC

…and how first measurements looked like

• NA38: first measurement of J/ suppression at the SPS, O-U collisions at 200 GeV/nucleon(1986), J/

periferiche centrali NJ/

cont. (2.7-3.5)

• J/ is suppressed (factor 2!) moving from peripheral towards central events

Peripheral events

Central events

Do we see a QGP effect in O-U collisions at SPS? Is this the end of the story ?

Early concerns…

…and different opinions…

… give us today the “flavour” of hot discussions held at that time

...showed that the story was not simple

• Are there any other effects, not related to colour screening, that may influence the yield of quarkonium states ?

• Is it possible to define a “reference” (i.e. unsuppressed) process in order to properly define quarkonium suppression ?

None of these questions (unfortunately!) has a trivial answer....

• Can the melting temperature(s) be determined ?

• Some basic topics/problems:

• What do we learn by looking at different quarkonium states?

Initial-state energy loss

• Energy loss of incident partons shifts x1

• √s of the parton-parton interaction changes (but not shadowing)

1

1

'

1 1

collN

gqxx q(g): fractional energy loss

• q =0.002 (small!) seems enough to reproduce Drell-Yan results • But a much larger (~factor 10) energy loss is required to reproduce large-xF J/ depletion from E866!

H.K.Woehri, “3 days of Quarkonium production...”, Palaiseau 2010

• New theoretical approaches (Peigne’, Arleo): coherent energy loss, may explain small effect in DY and large for charmonia

CMS explores the high pT region

67

(Maybe) we still see a hint of pT dependence of the suppression even in the pT range explored by CMS Good agreement with ALICE in spite of the different rapidity range (which anyway seems not to play a major role at high pT)

Centrality dep. in

pT

y

bins

Nuclear shadowing • Various parameterizations developed in the last ~10 years • Significant spread in the results, in particular for gluon PDFs • More recent analysis (EPS09), include uncertainty estimate

• Assuming a certain production approach (i.e. fixing the kinematics), the shadowing contribution to quarkonium production can be separated from other nuclear effects

K.Eskola et al., JHEP04(2009)065

(from C. Salgado)

Feed-down for bottomonium

Probing the QGP

One of the best way to study QGP is via probes, created early in the history of the collision, which are sensitive to the short-lived QGP phase

Ideal properties of a QGP probe

Production in elementary NN collisions under control

Not (or slightly) sensitive to the final-state hadronic phase

High sensitivity to the properties of the QGP phase

None of the probes proposed up to now (including heavy quarkonia!) actually satisfies all of the aforementioned criteria

So what makes heavy quarkonia so attractive ?

Interaction with cold nuclear matter under control

VACUUM

HADRONIC MATTER

QGP

Sequential suppression

71

Sequential suppression of the resonances

The quarkonium states can be characterized by • the binding energy • radius

More bound states smaller size

Debye screening condition r0 > D will occur at different T

state J/ c (2S)

Mass(GeV) 3.10 3.53 3.69

E (GeV) 0.64 0.20 0.05

ro(fm) 0.25 0.36 0.45

state (1S) (2S) (3S)

Mass(GeV) 9.46 10.0 10.36

E (GeV) 1.10 0.54 0.20

ro(fm) 0.28 0.56 0.78

(2S) J/ c

T<Tc

Tc

thermometer for the temperature reached in the HI collisions

(2S) J/ c

T~Tc

Tc

(2S) J/ c

T~1.1Tc

Tc

(2S) J/ c

T>>Tc

Tc

Alternative explanations?

Were indeed put forward very soon Two main categories 1) Cold nuclear matter effects

L

…that should occur already in pA collisions (which in turn can be used to estimate such effects)

2) Hot hadronic matter effects, via reactions like Expected to be sizeable for weakly bound states, such as (2S), but negligible for J/

XDDJ /

CNM effects: pA/dA collisions

One indeed observes a strong effect of cold nuclear matter on quarkonium: this led to a considerable effort of theory/experiment on the study of p-nucleus collisions Early studies concentrated on “nuclear absorption” as main effect, estimating effective quantities

NA50, pA 450 GeV

A significant reduction of the yield per NN collision is observed, usually parametrized by or abs

N.B.: J/pA/(A J/

pp ) is equivalent to RpA

ApppA absL

pppA Ae

~

J/ vs (2S): confirm break-up scenario

Less bound quarkonium states should be easier to break…. and indeed this is the case

At SPS energies

mb 0.98.3σψ'

abs

mb 0.54.5σ J/ψ

abs

Still, the dependence is weaker than the expected r2 one

state J/ c (2S)

Mass(GeV) 3.10 3.53 3.69

E (GeV) 0.64 0.20 0.05

ro(fm) 0.25 0.36 0.45

This fact is largely related to the quarkonium production mechanism

Quarkonium production Quarkonium is produced in two steps that can be factorized:

Production of the 𝑄𝑄 pair perturbative Evolution of 𝑄𝑄 pair into a bound quarkonium state non perturbative,

occurs on a relatively long timescale (several fm/c)

75

Color Singlet Model Color Evaporation M. NRQCD

Proposed soon after the J/ discovery 𝑄𝑄 pair is produced in a

color singlet state, with the same quantum numbers of the final quarkonium Unable to describe Tevatron data. However, recently NLO and NNLO corrections have been included to improve the agreement

𝑄𝑄 pair evolves in quarkonium if 𝑚𝑄𝑄 <mD

independently of its color and spin Probability to evolve into a certain quarkonium state depends by a constant F which is energy and process independent Works rather well, but no detail on the hadronization of the qq pair towards the bound state

Inclusive quarkonium production cross section is a sum of short distance coeff. and long distance matrix elements: This approach includes CSM as special case Charmonium can be produced also through the creation of a 𝑐𝑐 color octet state

/)/(ˆ J

n

n

ij

nQQOCJij

Kinematics

• By properly selecting the kinematics of the quarkonium states it is in principle possible to select events where resonance formation occurs inside (or outside) the nucleus

• A study vs xF is particularly relevant • Large-xF production resonance forms outside the nucleus • Small-xF production resonance forms inside the nucleus

• By varying the size of the target nucleus (i.e. performing systematic studies as a function of A) one can vary the thickness of nuclear matter L crossed by the cc pair (or the fully formed resonance)

p

c

c g

J/, c, ... c

c g

J/, c, ...

• The nucleus “sees” the cc in a (mainly) color octet state • Hadronization can take place outside the nucleus

Is nuclear absorption the whole story ?

77

Collection of results from many fixed target pA experiments, shown vs xF=2pL/s

Nuclear effects show a strong variation vs the kinematic variables

Very likely we observe a combination of several nuclear effects

lower s

higher s

J/

J/ vs (2S)

Fast (2S) as absorbed as J/!

E866

A cocktail with many ingredients

• Lots of interesting physics • Can we disentangle the various effects ? • Can we calculate them in a reliable way ?

• The break-up of the cc pair because of the interactions with CNM is an important effect, but other effects may also play a role

• Nuclear shadowing • Initial state energy loss • Final state energy loss • Intrinsic charm in the proton

Starting from the available fixed-target results (previous slide), do we reach a satisfactory understanding of what’s going on ?

(first systematic study, R.Vogt, Phys. Rev. C61(2000)035203)

Nuclear shadowing

valence quarks sea quarks gluons

79

PDF in nuclei are strongly modified with respect to those in a free nucleon

𝑓𝐴i(x,Q2)=Ri(A,x,Q2) x 𝑓𝑝i(x,Q2)

free proton PDF nPDF: PDF of proton in a nucleus

Various parameterizations available

Significant uncertainties for the gluon modifications, the more relevant for quarkonia production From enhancement to suppression, moving towards higher energy

SPS

RHIC LHC

Consequences

SPS Tevatron (FT) RHIC

Increasing √s From anti-shadowing to shadowing

At SPS, the “true” nuclear absorption cross section is larger than the “effective” one

Shadowing can be factorized: is nuclear absorption what remains ?

Is shadowing + absorption enough?

• Assume that the dominant effects are shadowing and cc breakup • cc break-up cross section should depend only on √sJ/-N

• Correct the results for shadowing (21 kinematics), using EKS98 • Even after correction, there is still a significant spread of the results at constant √sJ/-N

Effects different from shadowing and cc breakup are important

C. Lourenco, R. Vogt and H.K.Woehri, JHEP 02(2009) 014

NA3 – 200 GeV NA50 – 400/450 GeV E866 – 800 GeV HERA-B – 920 GeV

Another way to see it….

NA60, 2 sets of data taken with the same set-up, only changing Ebeam (good to reduce systematics)

Detect J/ at the same ylab same sN same nuclear absorption

Shadowing scales with x2 plot the results vs x2 and check if there is scaling

Yes: we can explain results in terms of shadowing+nuclear abs. No: evidence for other effects

x1=(mT/s) ey

x2=(mT/s) e-y

in a 21 kinematics

g

g

J/

Initial-state energy loss

The incoming parton producing the cc pair may undergo energy loss before the hard process takes place

Approach commonly followed: at each collision the parton looses a constant fraction of its energy ( x1 shift)

used in PDFs

used in cross section

number of NN

collisions ~ A1/3

fractional energy

loss

x1’ = x1 (1–q)N–1

Effect most important for fast partons large xF

Induces a smaller J/ yield reduced

Open charm affected too

Not clear if it completely solves the “puzzle” discussed in prev. slides

Moving to higher energies: dAu at RHIC

• Much larger √s at colliders, but: • Integrated luminosity smaller than at fixed target • Difficult to accelerate several different nuclei • Use one nucleus and select on impact parameter, but:

rT

b

pA: rT ~ b

rT’s b

dAu: due to the size of the deuteron <r>~2.5 fm the distribution of transverse positions are not very well represented by impact parameter

RHIC

• Is the situation becoming simpler at collider energies ?

Consequences

• Centrality classes do not probe completely unique regions and have a large amount of overlap

• Also shadowing estimates are less precise (need b-dependence, proportionality of effect with L usually assumed)

zdz

zrdzQxRNzrQxR

A

AA

iA

A

Li,0

,1,1,,, 22

,

(see S.R.Klein and R.Vogt, Phys. Rev. Lett. 91 (2003) 142301)

rT

L.A.LindenLevy, “3 days of Quarkonium production...”, Palaiseau 2010

J/ suppression in d-Au

• Regions corresponding to very different strength of shadowing effects have been studied (-2.2<y<-1.2, |y|<0.35, 1.2<y<2.2) good test of our understanding of the physics!

Fermi Motion

anti-shadowing

EMC effect

shadowing

x

forward y x~0.005 mid y x~0.03 backward y x~0.1

• In spite of RHIC starting its data taking in 2000, first high statistics dAu took place in 2008

A “selection” of PHENIX RAA results • Also at RHIC energies a superposition of shadowing+absorption is not satisfactory, compared to data

• In particular the relative suppression between peripheral and central events (RCP) is not reproduced

Shadowing linear case Absorption exponential case

None of them in agreement with data

Still, absorption scenario might work

McGlinchey, Frawley, Vogt, arXiv:1208.2667

Calculate the proper time spent by the cc pair in the nucleus

cc-N cross section (after correcting for shadowing) should depend on the size of the pair as it expands to a fully formed meson

=ZL/

with

“Large” data are in agreement with this simple hypothesis Other effects (energy loss?) not scaling with play a role in the small region

Data can be fitted with 1=7.2 mb, r0=0.16 fm and vcc=0

(2S) suppression in d-Au •Shadowing effects for J/ and (2S) should be very similar •At RHIC energy the final meson state should form outside the nucleus absorption effects expected to be similar

• In contrast to these expectations, much stronger (2S) suppression!

Are we observing “hadronic comovers”?

Charmonia in cold nuclear matter: what did we learn?

Cold nuclear matter effects (both initial and final state) strongly modify charmonium spectra

Physics interesting in its own but also for the understanding of what is observed in AA collisions

Main features are understood, thanks to the large amount of measurements covering large phase space regions

Quantitative description still lacking Large uncertanties on shadowing Energy loss effects Modelling of cc formation vs time still rather simplistic

Consequence: extrapolation of CNM to AA collisions not completely quantitative mostly phenomenological approaches

Still, charmonium/bottomonium AA data contain very rich physics and represent a unique set of observables for QGP studies!

…as we are going to see in the next slides !

Still on resolution

0s X

xzβcp

MeV 21.2θ

B0.3

p

L

ε

5N

720

p

δp2

• Multiple scattering is dominant for low-momentum muons

• For high-momentum muons the resolution is dominated by tracking accuracy (p/p proportional to p)

• The variance s of the angular distribution is proportional to 1/p

• At mµµ ~ 1 GeV, track matching allows a strong improvement in the invariant mass resolution

• At mµµ ~ 3 GeV (J/) the contribution of multiple scattering is less important but not negligible

• In a muon spectrometer the invariant mass resolution has therefore two components, the one on multiple scattering is important at low masses

What is (heavy) quarkonium ?

92

Quarkonium is a bound state of 𝑞 and 𝑞

q

q

Particular interest in states having 𝑚𝑞𝑞 < 2𝑚𝐷(𝑚𝐵)

Charmonium (𝑐𝑐 ) family Bottomonium (𝑏𝑏 ) family

Several quarkonium states exists, distinguished by their quantum numbers (JPC)

M(J/) = 300 MeV

1 nb

1 pb

M (GeV)

p-U at 20-30 GeV

Christensen et al.,

Phys. Rev. D8 (73) 2016

?

• (2S) : difficult detection, visible as a shoulder of J/

Cattiva risoluzione in massa ?

NA51

E605

Physics and mass resolution

Sources of heavy quarkonia

94

Quarkonium production can proceed:

• directly in the interaction of the initial partons • via the decay of heavier hadrons (feed-down)

For J/ (at CDF/LHC energies) the contributing mechanisms are:

Direct production

Feed-down from higher charmonium states: ~ 8% from (2S), ~25% from c

B decay contribution is pT dependent ~10% at pT~1.5GeV/c

Pro

mpt

Non-p

rom

pt

B-decay component “easier” to separate displaced production

Direct 60%

B decay 10%

Feed Down 30%

Low pT

Invariant mass spectrum • Having an experiment which measures lepton pairs it is then easy to build the invariant mass spectrum of the pairs and access in this way many physics processes

)2(

))((//

eeeeee

eeeeJJ

pppppp

pppppp

)cos(22 22

/ eeeeeeeJ ppEEmm

one easily gets

Let’s consider the 2-body decay J/e+e-

(it is necessary to measure momentum and opening angle of leptons)

At high energy E~p, me ~0 2

sin2/

ee

eeJ EEm

A look to real data • The physics ingredients listed in the previous slide constitutes our “signals” see a “real data” example from NA60

• Unfortunately also background sources are present, need to be identified and subtracted for a physics analysis to be carried out

Dilepton experiments combinatorial background

In-In @ 158 GeV/nucleone

Combinatorial background

• Mainly due to uncorrelated pion/kaon decays

• How minimizing it? Hadronic absorber close to the target region (not too much!)

• Subtraction techniques • An experiment studying muon/electron pairs detects not only opposite-sign dileptons but also like-sign (e.g. ++ e -- ) • Like-sign pairs are due to combinatorial background

• Let’s suppose that the apparatus has the same acceptance (in a certain kinematic range) for + and -

A++=A--=A+-

• Let’s suppose to produce • N+ (N-) positive (negative) mesons per event • P(N+) (P(N-)) probability of producing N+(N-) decay muons • P(N+,N-) probability of produding N+ positive and N- negative

(Let’s finally suppose that there are no strong correlations between charges of particle decaying to muons (true at high multiplicity))

Estimating combinatorial bckgr.

•With the hypothesis from previous slides

NNA

dNNN

NPAN 2

22

)1()(

NNA

dNNN

NPAN 2

22

)1()(

and

ANNdNdNNNNNPAN ),(

If meson multiplicity is Poissoniand and therefore 22 NNN

AA

ANNN 2 NNN 2

= 1 when acceptance is symmetric

since )()(),( NPNPNNP (no charge correlations)

At SPS energy This is indeed a powerful approach when statistics is large and was widely used e.g. at SPS energies ….

NA50, PbPb s = 17.3 AGeV

..where by the way the contribution of combinatorial background was vanishing at around quarkonium mass…

At LHC energy ..but it is less suitable when the significance of the signal is small

In this situation, the so-called “mixed event” approach proves to be more powerful

Details may vary, but in general

1)Bin the events according to their characteristics (kinematic, vertex position, centrality) 2)Build pools of /e of different charge 3)Combine /e from pools to build “mixed event” opposite sign di-muon/-electron spectrum 4)Normalize to measured opposite sign spectrum (various prescriptions possible)

With this method, statistical uncertainties are reduced

From the spectrum to NJ/ After subtraction of the combinatorial background (or after having “fixed” its contribution to the spectrum) the number of charmonium decay events can be extracted…..

…via a cocktail fit, knowing the expected mass shape of the various processes

J/

’

DY

Background

Charm

… or simply parameterizing the non-resonant part via a phenomenological function

NA60 InIn s=17.3 A GeV

A (slightly) closer look to experiments: CMS

102

“Global” muons reconstructed with information from inner tracker and muon stations

Further muon ID based on track quality (2, # hits,…) Magnetic field and material limit minimum momentum for muon

detection pT cut for J/

Tracker pT resolution: 1-2% up to pT~100 GeV/c Separation of quarkonium states Displaced tracks for heavy-flavour measurements

From NJ/ to suppression (1)

High temperature should indeed induce a suppression of the charmonia and bottomonia states

How can we quantify the suppression ?

Low energy (SPS) Normalize the charmonia yield to the Drell-Yan dileptons

g

g

c

c

J/

q

q

*

+

-

+

-

Advantages Same final state, DY is insensitive to QGP Cancellation of syst. uncertainties

Drawbacks Different initial state (quark vs gluons) High-mass DY has small statistics

From NJ/ to suppression (2)

At RHIC, LHC Drell-Yan is no more “visible” in the dilepton mass spectrum overwhelmed by semi-leptonic decays of charm/beauty pairs

Solution: directly normalize to elementary collisions (pp), via nuclear modification factor RAA

𝑅𝐴𝐴 = 𝑑𝑁𝑃

𝐴𝐴

𝑁𝐶𝑜𝑙𝑙

𝑑𝑁𝑃

𝑁𝑁

RAA<1 suppression RAA>1 enhancement

Advantages same process in nuclear environment and in vacuum Drawbacks Systematics more difficult to handle (no cancellations)

An ideal normalization would be the open heavy quark yield However, this is not straightfoward (see later)

A (slightly) closer look to experiments: ALICE

105

Main difference with respect to CMS: PID over a large pT range, down to low pT (~0.1 GeV/c)

TPC as main tracker slower detector, lower luminosity

“Intermediate” situation for the forward muon arm Faster detectors, can stand higher luminosity

Experiments From fixed target at the SPS

(muons only)…

NA60

to RHIC collider (muons+electrons)…

Experiments

…to the LHC (electrons+muons)

CMS (high pT)

ALICE dedicated HI experiment CMS+ATLAS mainly pp, but very good capability for charmonia and bottomonia in HI

Rapidity dependence

108

Rather pronounced in ALICE, and evident in the forward region (~40% decrease in RAA in 2.5<y<4) More difficult to conclude between mid- and forward-rapidity PHENIX-like ??

Shadowing estimate

(EPS09, nDSg)

Compatible with central,

NOT with forward y

More general CNM issue

ALICE, focus on low-pT J/

109

Electron analysis: background subtracted with event mixing Signal extraction by event

counting

|y|<0.9

Muon analysis: fit to the invariant mass spectra signal extraction by

integrating the Crystal Ball line shape

What did we learn ?

110

26 years after first suppression prediction, this is observed also in the bottomonium sector with a very good accuracy

RAA vs binding energy qualitatively interesting: can different pT coverage be seen as a way to “kill” recombination ?