Medium and High pT Direct Photons from PHENIX at RHIC A.Bazilevsky For the PHENIX Collaboration Winter Workshop on Nuclear Dynamics February 3-10, 2013.

1

Medium and High pT Direct Photons from PHENIX at RHIC

A.BazilevskyFor the PHENIX Collaboration

Winter Workshop on Nuclear Dynamics February 3-10, 2013

Squaw Valley, CA

2

Direct Photons in (polarized) pp

Quark-Gluon Compton scattering is the dominant source of high pT photons in pp collisions (q-qbar annihilation – 10-15%)

Access to gluon (polarized) distributions through factorized pQCD

No fragmentation involved -> access to Sivers function in transversely polarized protons (correlation between proton spin and parton kT)

TMD factorization?

Sivers function universality?

… Also fragmentation photons

See Spin talks by C.McKinney and O.Eyser

3

Direct photons in HIC

Initial Condition:hard

QGP:thermal Jet fragmentationBremsstrahlungJet-photon conversion

Hadron gas:thermal

Time

Photons emitted at all stages of the system evolution

With different characteristic pT spectrum

Once produced photons leave the medium unmodified

Carry dynamical information of the state

Initial state

Init. temperature, thermalization time

Modification of fragmentation (in γ-jet)

…

Turbide, Rapp, Gale,Phys. Rev. C 69 (014903), 2004

4

PHENIX detector for photon measurements

Fine segmented EMCal for γ+RICH&Tracking&Magnet for e+ e-

BBC&ZDC for event characterizationBBC&RxNP for reaction plane

High rate DAQ: 5-7 kHz

TriggerMB (collision) with BBCRare with EMCal (γ), EMCal&RICH (e)

e+e-

pg

5

Direct γ reconstruction: higher pT

p0-tagging (pp and dA):

Statistical subtraction:

80 12 pT(GeV/c)4

pp

RAuAu

Direct γ = Inclusive γ – Hadron_Decay γ

6

Direct γ reconstruction: lower pT

• Source of real photons is also a source of virtual photons decaying into e+e-

• Focus on the mass region where p0 contribution dies out: Mee >mπ0

• For Mee<<pT : M<300 MeV/c2

– qq ->* contribution is small– Mainly from internal conversion of

photons

• Can be converted to real photon yield using Kroll-Wada formula (for Dalitz decay spectra)

One parameter fit: (1-r)fc + r fd

fc: cocktail calc., fd: direct photon calc.

SdNMM

m

M

m

dM

Nd

eeee

e

ee

e

ee

121

41

3

22

2

2

22

q

g*

g q

e+

e-

S~1 for Mee << pT

PRL 104,132301 (2010)

Direct γ = Inclusive γ – Hadron_Decay γ See talk by S.Rolnick

7

pp

Measured up to pT=25 GeV/c

The pQCD NLO calculation is in good agreement with data

Important reference for HI measurements to study hot and cold nuclear effects

PRD 86, 072008 (2012)

The pT range in the previous report. (PRL 2007)

8

pp: scaling

• Plotting cross-sections in p-p and p-pbar from various experiments against xT = 2pT/√s

• Hard scattering process should ~scale with xT (if no αs, PDF evolution, etc.)

• Scale yields by (√s)n

• PHENIX data includes virtual photon results in low pT(1<pT<5GeV/c)

• n=4.5 makes a universal line

• n=4 for pure gluon exchange with no evolution of αs, PDF, etc.

PRD 86, 072008 (2012)

9

pp: event shape

Checked the event shape with an isolation cut

Generally the theory calculation agrees with dataDeviation from theory at low pT?

How reliable is theory at low pT?

At high pT most direct photons are isolatedDecay photons are not isolated (as expected)

Econe < 0.1E

PRD 86, 072008 (2012)

10

Application to proton spin structure studies

Clean access to ΔG

But needs more luminosity and larger acceptance (-> RHIC upgrade, sPHENIX)

See talk by C.McKinney

See talk by O.Eyser

Projected stat. sensitivity3.0<|η|<3.8Includes correction for background from π0, η

Gives access to Sivers function (correlation between proton spin and parton kT)

11

AuAu: RAAPRL 109, 152302 (2012)

L in p+pNcoll

RAA~1 : Consistent with binary scaling of p+p.

Minimum bias

0-5%

60-92%

… Though many mechanisms may affect it (and/or compensate each other): Isospin difference, nPDF, CroninSuppression of fragmentation photonJet-photon conversion, Bremsstrahlung in QGP, etc.

12

Also CuCu-200GeV and AuAu-62GeV

Cu+Cu at √sNN = 200 GeV

RAA~1 : Consistent with binary scaling of p+p.

13

AuAu: RAA

Initial state effects (IS) include isospin and nuclear PDF, consistent with data

Final state effects (FS) include suppression of jet fragmentation photons and photons from jet-plasma interaction, consistent with data

Another model with both IS and FS disagrees with data

JHEP1104, 055

PRC77, 024909arXiv:0904.2184

PLB669, 337

PRL 109, 152302 (2012)

Need for more precise data

14

dAu

Rd+Au is consistent with 1.It suggests little or no nuclear effect within our uncertainty.Consistent with theory calculations with CNM effects

arXiv:1208.1234Theory calculations fromI.Vitev et al., PLB669, 337 (2008)

15

dAuWe have x30 statistics in Run8 compared to published Run3. Will help to constrain the gluon PDF further.

arXiv:1012.1178

RHIC range

16

v2

annihilationComptonscattering

Medium induced(inc.energy loss)

jet

jet fragment photon v2 > 0

v2 > 0

v2 < 0Hard scattering photons: v2~0

• V2 depends on production processS. Turbide, C. Gale and R.J. Fries, PRL 96, 032303 (2006)

• V2 is sensitive to thermalization timeR. Chatterjee and D. K. Srivastava PRC 79, 021901 (2009)

More details in talk by S.Rolnick

17

v2

At high pT, v2~0. It is consistent with the hard scattering source. At low pT (thermal region), an unexpectedly large v2 is observed!

It depends on the formation timeTheory predictions are a bit small.

PRL 109, 122302 (2012)

MB (0-92%) 0-20% 20-40%

calculation from PRC 79, 021901(R)

More discussion on low pT v2 in talk by S.Rolnick

18

PHENIX upgrade for photons

3.1<h<3.8

To be ready for Run-14/15

Initial condition for HIGluon Distribution in CNM at Low-x

Direct photons (h, 0 suppression + isolation cut)

MPCEX

A combined charged particle tracker and EM preshower detector – dual gain readout allows sensitivity to MIPs and full energy EM showers.

• Direct photons• p0 reconstruction out to >80GeV• Charged track identification

MPC

MPC-EX

More details in talk by S.Cambell

19

PHENIX -> sPHENIX

A plan to upgrade the detector.

Large acceptance with high /0 separation.More tracking layers, hope eID is still powerful.Hadron calorimeter for +Jet analysis.

High Rate capability and large acceptance : 106 jets per year above 30 GeV/c104 direct per year above 20 GeV/c

More details in talk by D.Morrison

20

Summary• p+p reference measurements are on the universal curve of world’s

measurements; in agreement with NLO pQCD

• dAu: RdA ~1; expected nuclear modification within current uncertainites

• AuAu: No nuclear modification (RAA~1) in high pT photon (> 5 GeV/c)

• v2~0 at pT>4 GeV/c: consistent with prompt (hard) photons

• A challenging probe, both experimentally and theoretically Need for more precise measurements in wider kin. range with a variety of observables (RAA, v2, v3, …, HBD …) Short and long term PHENIX upgrade plan

γ for SpinBy C.McKinney and O.Eyser

Low pT and v2By S.Rolnick

γ-h correlationBy A.Hanks

PHENIX UpgradeBy S.Campbell and D.Morrison

21

Backup

22

AuAu

• Long standing issue of direct photons RAA at high pT

23

AuAu

PRL 109,152302 (2012)

24

Low pT -> thermal arXiv:1208.1234

• g fraction = Yielddirect / Yieldinclusive

• Lines are NLO pQCD calculation with scales μ= 0.5pT, 1.0pT, 2.0pT

• Largest excess above pQCD is seen at Au+Au

• Nuclear effect measured in d+Au does not explain the photons in Au+Au

RAA>1 at pT<3 GeV/c, whereas RdA~1

Medium effects (additional source of direct photons) in AuAu

25

Low pT -> thermal

Au+Au

PRL 104,132301 (2010)

First information about the temperature of the system averaged over the space-time evolution of the collision

From fit to exp + ncoll scaled pp:

Tave = 221 19stat 19syst MeV (MB)

Small (if any) centrality dependence

NLO pQCD consistent with p+p down to pT=1 GeV/c

Excess of photons (with 1<pT <3 GeV/c) in Au+Au beyond the Ncoll

scaled p+p yield.

Interpreted as thermal radiation emitted by the medium

26

Initial temperature & formation time

Tc~170MeV from lattice QCD

PRC 81, 034911 (2010)

All above the critical temperature Tinit depends on 0 assumption

Need observable to constrain initial conditions

Tini = 300 to 600 MeV t0 = 0.15 to 0.5 fm/c

27

dA

28

Initial kT broadening or recombination?• Recombination model claims that the Cronin effect in hadron production is

built up by recombination– e.g. R. Hwa, Eur.Phys.J.C43:233(2005)– Cronin effect in direct photon production should be smaller than one in p0

• Within quoted errors, the effect is same for p0 and photon production

p0 RAA in d+Au at 200GeV. PRL91, 072303 (2003)

29

• Bremsstrahlung from hard scattered partons in medium (Jet in-medium bremsstrahlung)

• Compton scattering of hard scattered and thermal partons (Jet-photon conversion)

30

Also CuCu, AuAu-62GeV

Cu+Cu at √sNN = 200 GeV

31

Au-Au vs Cu-Cu

Could d+d collisions help? (One could tag pp, pn, nn collisions!)

At 62GeV there is no experimental difficulty

The isospin effect should be (almost) independent of centrality

Little overlap with Cu+Cu, but the two are consistent within errors

32

Is it (only) an isospin effect?• Taking for example, the isospin effect: Direct photon cross-sections for p+p, p+n and n+n are different because of different charge contents ( eq

2)

• Effect can be estimated from NLO pQCD calclation of p+p, p+n and n+n– In low pT, quarks are from gluon split no difference between n and p– At high pT, contribution of constituent quarks manifests

• Minimum bias Au+Au can be calculated by:

)Z)-(A Z)-2Z(A (Z)(1/A /N nn2

pnpp22

collAA

(sAA/Ncoll)/spp vs pT (sAA/Ncoll)/spp vs xT

Same suppression will be seen in lower pT at sNN=62.4GeV

TS, INPC07, arXiv.org:0708.4265

33

STAR

STAR, Phys.Rev.C81,064904(2010)

• Excess in d+Au?– No exponential excess

• High-pT direct photon results from PHENIX and STAR– d+Au

• Agree with TAB scaled pQCD

• consistent with PHENIX and STAR– p+p

• Agree with pQCD and PHENIX

• Low-pT direct photon – No publication data at STAR

34

Mee

Phys. Rev. Lett. 104, 132301 (2010)

35

v2

12

.2.

2

R

vvRv

BGincdir

Consistency between EMCal photons and internal conversion photons (at low pT)

36

v2

• High pT (pT>5 GeV/c)– v2 ~ 0 (independent of

centrality)– Consistent with STAR

results within large error.

• Low pT (pT<3 GeV/c)– Inconclusive centrality

dependence

STAR, arXiv:1008.4894

PHENIX, arXiv:1105.4126

Cent20-40%Cent0-20%

Cent10-40%

37

Direct photon v2

PRL 109, 122302 (2012)

Au+Au √sNN = 200 GeVminimum bias

• Large v2 at pT < 4 GeV/c where thermal photons dominate

• v2 consistent with 0 at high pT where prompt photons domininate

Very surprising result: large v2 implies late emission whereas thermal radiation implies early emission

Models have difficulties in reproducing simultaneously yield and v2 of photons

38

Direct photon v2

pT (GeV/c)

dir. v

2

Large v2 of low pT thermal photon

200GeV Au+Au 0-20%

R. Chatterjee and D. K. SrivastavaPRC 79, 021901(R) (2009)PRL96, 202302 (2006)

Several models have failed in v2 magnitude with similar shape

• Thermal photon in quark matter– v2>0 at low pT

– v2~0 at high pT

• Thermalization time t0

– Early (smaller v2)

– Late (larger v2)

• Constrain t0

– Measure v2 at low pT

Hydro after t0

v2

39

39

0904.2184

Hydro region

pQCD region

Needs *2-3 smaller errors to become decisive

40

This fits to data well, but..• Large flow can not be produced in partonic phase, but could be in

hadron gas phase

• This model changed ingredients of photon spectra drastically!– We realized the importance of the data…

thermal + prim. g

van Hees, Gale, Rapp, PRC84, 054906 (2011)

41

v2: theory comparison0-20% Phys. Rev. C 84, 054906 (2011)

H. van Hees, C. Gale, R. Rapp

• Important features/differences from hydrodynamic expansion– Larger equilibrium hadronic rates– Hadronic phase includes meson-chemical potentials– Hadronic phase lasts longer (smaller Tfo and larger Tch)– Elliptic flow builds up faster

• Thermal radiation dominated by hadronic phase• Is the QGP window closed?

42

v2

43

FF shape modification

43

xE = -pTh/pT

g cos(Df)~zT = -x ln xE - MLLA variable Universal scaling in pp Enhancement at very low zT Suppression at large zT

- pQCD photons provide a reasonable estimate for the energy of recoil jet (q or g);

- Measured medium FF is softer compared to vacuum fragmentation;

- Further studies with jets tagged by strangess (gluon / quark jet separation) follow