1. Heavy Ion Measurements with the ATLAS Detector at the LHC Brian A. Cole, Columbia University June 28, 2007

2. 2 LHC Heavy Ion Program: Key Questions What is the mechanism for initial particle production at the LHC? –Production from a saturated initial state? How rapidly do produced particles thermalize or isotropize, what is the mechanism? –Faster than RHIC, slower ? How do high-energy quarks and gluons interact in the quark gluon plasma? –What is the response of the medium? What is the screening length of the QGP? What are the quasi-particles of the QGP? How does the QGP hadronize?

3. 3 LHC Heavy Ions Program: Key Questions

4. 4

5. 5 LHC Physics Summary High p T Low x Parton Density RHIC Collectivity ??

6. 6 Jet Tomography At RHIC, studied via leading hadrons –Statistics suffer from frag. function  rates –Quenching  geometric bias –No direct measure of frag. function. At LHC: –Full jets, high p T, large rates, b jets, di-jet,  -jet  Precision jet tomography

7. 7 Parton Showers, Hard Radiation @ LHC Copious hard radiation in high Q 2 final- state parton showers,  F ~ 1/k T Both an opportunity and a challenge –Understanding jet quenching more difficult –Potentially: time-dependent probe of medium Resolving hard radiation in jets a must!

8. 8 The ATLAS Central Detector Inner tracking, EM and Hadronic calorimeters, external muon spectrometers

9. 9 ATLAS from the Inside

10. 10 ATLAS Zero Degree Calorimeter Test beam @ CERN 10.06 ZDC Prototype @ CERN 10.06 Experiment Simulation p+p events, w/ precision EM module TAN region, z=140m

11. 11 ATLAS Acceptance

12. 12 ATLAS Acceptance Si Tracking Muon spectrometer EM Calorimeter Hadr Calorimeter ,  ’ ,  0, isolated  Jets Bulk observables ZDC s 

13. 13 ATLAS Acceptance ,  ’ ,  0, isolated  Jets Bulk observables 

14. 14 ATLAS Inner Tracker & Tracking 3 layers Si pixel 8 layers Si strip TR tracker “Vanilla” ATLAS track reconstruction –But with only Si detectors Tuned parameters but –No  dependence to tracking cuts (yet) –No verification with calorimeter clusters (here) |  |<0.6

15. 15 ATLAS Calorimetery EM Long. Segmentation Hadronic Barrel Hadronic EndCap EM EndCap EM Barrel Forward

16. 16 ATLAS Heavy Ion: Primary Goals Measure dn chg /d , dE T /d  (total+EM) –Characterize gross properties of initial state. –Test saturation predictions Measure charged, inclusive ,  0 elliptic flow –Probe early collective motion of (s/t/w)QGP Measure jets, jet fragmentation,  -Jet, di-jet, … –Precision tomography of QGP & its properties –Medium response to passage of quenched jet Measure Upsilon production via  +  - –Probe Debye screening in medium Study low x hard processes in p-p, p-A –Study factorization violations, BFKL, saturation

17. 17 Multiplicity from Si Hits Count clusters in each of 3 pixel layers Correction = Apply to simulated events (single) HIJING, Pb+Pb, b = 2.3fm b = 10.7fm

18. 18 Charged Multiplicity from Tracklets 1. Truth tracks (black) 2. “B-Layer” Hits 3. Layer 1 Hits 4. Matched Tracklets η Φ Pixel “tracklets” Histogram (yellow) – Hijing Points (black) – Raw tracklets (no corrections)

19. 19 Charged Multiplicity from Tracklets Tracklets provide a good estimate of multiplicity –< 20% correction dn/d  resolution ~ 2% –except for most peripheral events

20. 20 E T Measurement dEt/d  HIJING b=2.3, corrected RHIC:  E T  /particle ~ constant Reproduced by HIJING E T in ATLAS (full simulation)

21. 21 Reaction Plane with Calorimeters Flow imposed on HIJING events via  shift –a la Poskanzer and Voloshin, PRC 58 (1998). –Parameterization of RHIC v 2 (N part, , p T ) Beware suppressed zero Presampler Layer 1Layer 2Layer 3 ϕϕϕϕ ϕϕϕϕ ηηη η Barrel calorimeter only

22. 22 Reaction Plane with Calorimeters Flow imposed on HIJING events via  shift –a la Poskanzer and Voloshin, PRC 58 (1998). –Parameterization of RHIC v 2 (Npart, p T ) Characterize  RP resolution with v 2 correction factor v 2 correction factor Comparison to true  RP Comparison of subevents

23. 23 Reaction Plane w/ Tracks As above, HIJING +  shift w/ parameterized RHIC v 2 (Npart, p T ) Use reconstructed tracks p T > 0.5 GeV/c, |  |<2.5 Reaction plane resolution via v 2 correction factor

24. 24 Jets in A+A Jets from PYTHIA in 0.1x0.1 (logical) towers

25. 25 Jets in A+A merged with b = 2 fm Pb+Pb event (HIJING) Jets from PYTHIA in 0.1x0.1 (logical) towers

26. 26 Jet Reconstruction: E T Resolution Pythia di-jet events with 35 < E T < 280 –Merged (post GEANT) into b = 2 fm HIJING events. Reconstructed w/ R=0.4 seeded cone algorithm –Seed: E T > 5 GeV in  = 0.1x0.1 tower Compared to R=0.4 seeded cone algorithm on Pythia final-state particles. 

27. 27 Jet Reconstruction: b dependence Pythia + HIJING performance vs b –R = 0.4 seeded cone jet algorithm –Here, Pythia jets in 70 < E T < 90 GeV –Position resolution –Energy resolution Smooth evolution with centrality By b =10 (N part = 100) reach  p-p performance. RMS  RMS  E/E Pythia Jets: 70 < ET < 90 GeV

28. 28 Jet Fragmentation Observables

29. 29 k T Jet Reconstruction k T jet algorithm has several advantages –Unseeded (better QCD predictability) –Explicitly accounts for angular ordered parton showers –Adapts to distorted (non-conical) jet shapes Shamelessly borrowed from talk by W. Holzmann

30. 30 k T Jet Reconstruction k T jet algorithm has several advantages –Unseeded (better QCD predictability) –Explicitly accounts for angular ordered parton showers –Adapts to distorted (non-conical) jet shapes With algorithmic optimization by Cacciari, becomes feasible in Pb+Pb (faster than cone)

31. 31 k T Jet Reconstruction (2) Cacciari: –Use K T algorithm w/o subtraction. –Use fake jets to estimate background, subtract. ATLAS: –Use jet using  = 0.1  0.1 towers to distinguish real & fake jets. 3 4 1 2 3 4 2 1 Central Pb+Pb event + q T = 140 GeV Pythia, EM energy only

32. 32 k T Jet Reconstruction (3) Very preliminary K T algorithm with “R=0.4” –E T max /  ET  cut at avg. + 1  1 st study of performance of fast k T algorithm in Pb+Pb But a crucial proof-of-principle showing the method works Avg = 2.0 RMS = 0.9 E T max /  E T  # jets

33. 33  Detection: Intrinsic Performance Good mass resolution Large acceptance Loss of efficiency near  ~0 due to gaps, supports

34. 34 Upsilon Measurement in (Central) Pb+Pb Mass distributions w/ background muons Upsilon spectrum w/ Pb+Pb resolution (|  |<2) |  |<1 |  |<2

35. 35 Summary Heavy ion physics is integral part of ATLAS ATLAS provides unique capabilities –(e.g.) Highly segmented (longitudinal and transverse) EM calorimeter covering >6 units of pseudo-rapidity. What I covered –Event characterization: multiplicity, E T –Reaction plane measurement and v 2 –Jet reconstruction: cone and K T –Upsilon measurements What I didn’t have time to show you –Ultra-peripheral, low-x via  +A –Gamma-jet,  /  0 /  separation in EM calorimeter –Z-jet, jet-jet, … –B-tagged jets (future work)

36. 36 ATLAS HI Working Group A. Ajitanand 10, A. Angerami 3, G. Atoian 11, M. Baker 1, P. Chung 10, B. Cole 3, R. Debbe 1, A. Denisov 5, J. Dolejsi 2, N. Grau 3, J. Hill 7, W. Holzmann 3, V. Issakov 11, J. Jia 10, H. Kasper 11, R. Lacey 10, A. Lebedev 7, M. Leltchouk 3, A. Moraes 1, R. Nouicer 1, A. Olszewski 6, A. Poblaguev 11, V. Pozdnyakov 8, M. Rosati 7, L. Rosselet 4, M. Spousta 2, P. Steinberg 1, H. Takai 1, S. Timoshenko 9, B. Toczek 6, A. Trzupek 6, F. Videbaek 1, S. White 1, B. Wosiek 6, K. Wozniak 6, M. Zeller 11 1 Brookhaven National Laboratory, USA 2 Charles University, Prague 3 Columbia Unversity, Nevis Laboratories, USA 4 University of Geneva, Switzerland 5 IHEP, Russia 6 IFJ PAN, Krakow, Poland 7 Iowa State University, USA 8 JINR, Dubna, Russia 9 MePHI, Moscow, Russia 10 Chemistry Department, Stony Brook University, USA 11 Yale University, USA

37. 37 J/  and Upsilon Rates  J/ 

38. 38 Jet Background EM Calorimeter Long. Segmentation Jet Back ground All too wide for single photons Segmentation of first EM sampling layer so fine that heavy ion background is ~ negligible –  N chg + N   < 1,  E T  ~ 30 MeV Fine   rejection of neutral hadron decays Clean 1 st sampling  prompt  isolation

39. 39  /  0 Separation w/ EM First Layer Can use the fine segmentation of 1 st EM layer to reject  0 and  decay photons. Run 10040, Ev. 1 Photon Cl.  =1.93, Cl. E T =17.5 GeV 17.5 GeV  Run 10083, Ev. 5 Pizero (2  ) Cl.  =-0.52, Cl. E T =28.7 GeV 29 GeV  0 Energy/strip

40. 40  /  0 Separation w/ EM First Layer (2) Left – fraction of energy outside shower “core” in strips Right – energy of second maximum in strips Compare ,  0

41. 41  /  0 Separation w/ EM First Layer (3)  0 rejection for 90%  efficiency vs  and p T –  rejection a factor of ~ 4 larger –  unaffected by Pb+Pb background Rejection modest compared to isolation (> 10 2 ) –But, any improvement crucial –Allows direct, high statistics measurement of bkgd

42. 42 Jet Tomography At RHIC, studied via leading hadrons –Statistics suffer from frag. function  rates –Quenching  geometric bias –No direct measure of frag. function. At LHC: –Full jets, high p T, large rates, b jets, di-jet,  -jet  Precision jet tomography

43. 43 ATLAS: Gamma-Jet Pythia  + jet (75 GeV) superimposed on b=4 fm HIJING Pb+Pb event, full GEANT   Jet Gamma

44. 44 ATLAS: Gamma-Jet Pythia  + jet (75 GeV) superimposed on b=4 fm HIJING Pb+Pb event, full GEANT   Background subtracted Jet Gamma

45. 45 ATLAS: Gamma-Jet, EM 1 st Layer Gamma 1 st layer unaffected by Pb+Pb background  isolation w/ 1 st layer ~ unaffected by Pb+Pb Zoom in on barrel EM calorimeter 1 st sampling layer

46. 46 Low-x Physics w/ ZDC 12 π 0 acceptance Log 10 (x 2 ) p T (GeV) ZDC w/ precision EM module measures semi-hard  0, , , … production at x ~ 10 -6 in p-p and p-A collisions Correlate with mid-rapidity jets

47. 47 ZDC  0, , , … Reconstruction In p-p and p-A, position resolution of ZDC EM module allows clean measurement of  0, , , … –Huge benefit to low-x physics program Obtained with cut on total energy, E > 200 GeV Very little background from non- vertex sources

48. 48 ATLAS Tracking Fake track fraction High fake track rates for |  | > 2. –Increasing hit density –Material in Si tracker  More work needed  e.g. tighter cuts @ larger  Tracking efficiency vs , pT for properly reconstructed tracks –fake tracks DO NOT artificially increase eff. –Efficiency ~ flat w/ p T