Heavy Ion Physics with the ATLAS Detector

1
Heavy Ion Physics with the ATLAS Detector
Barbara Wosiek Barbara.Wosiek_at_ifj.edu.pl Institute
of Nuclear Physics, Kraków, Poland
For the ATLAS Heavy Ion Group
S. Aronson, K. Assamagan, B. Cole, M. Dobbs, J.
Dolejsi, H. Gordon, F. Gianotti, S. Kabana, M.
Levine, F. Marroquin, J. Nagle, P. Nevski, A.
Olszewski, L. Rosselet, P. Sawicki, H. Takai, S.
Tapprogge, A. Trzupek, M.A.B. Vale, S. White, R.
Witt, B. Wosiek, K. Wozniak and
2
Outline of the Talk
INTRODUCTION

• Why Heavy Ions at the LHC?
• Why ATLAS as a Detector for Heavy Ions?

ATLAS PERFORMANCE FOR HEAVY ION PHYSICS

• Monte Carlo Simulations
• Detector Occupancies
• Global Measurements
• Event Characterization
• Tracking with ATLAS ID(Si)
• B-tagging
• Quarkonia Studies
• JET PHYSICS ? Ketevis talk

3
Heavy Ions at the LHC
Study of QCD matter at extremely high energy
densities and vanishing baryon chemical
potential.

• deconfinement
• restoration of the chiral symmetry,
• physics of parton densities close to saturation

RHIC LHC 200 5500 GeV

• Initial energy density about 5 times higher than
  at RHIC.
• Lifetime of a hot dense matter much longer
• 10-15 fm/c at LHC as compared to 1.5-4 fm/c
  at RHIC
• Access to truly hard probes with sufficiently
  high rates
• pT gt 100 GeV/c (at RHIC pT ? 20
  GeV/c)
• copious production of b and c quarks

4
Heavy Ions at the LHC

• Quantitative studies of a QGP properties
• Hot/dense matter effects should dominate over
• initial and final state effects.
• Studies facilitated by many hard probes.

LHC HI Phase I
Pb Pb E2.75 TeV/beam Lmax 11027
cm-2 s-1 Interaction Rate 8 kHz Exploratory
run of a few days in 2007 Extended run in 2008 (1
nb-1)
LHC HI Phase II and later

• p A collisions (benchmarking nuclear effects)
• Possible lighter ion species 115 In, 84Kr,
  40Ar, 16O

5
Heavy Ion Physics with ATLAS Detector
ATLAS interest in heavy ion physics was activated
by highlights from RHIC experiments!
Hot/dense Nuclear Matter Diagnostics

• Suppression of high pT particles
• Disappearance of back-to-back high pT jet
  correlations
• Huge azimuthal asymmetry at high pT

ATLAS is an excellent detector for high pT
physics and jet studies
6
ATLAS as a Heavy Ion Detector

• High Resolution Calorimeters
• Hermetic coverage up to ? lt 4.9
• Fine granularity (with longitudinal segmentation)

High pT probes

• Large Acceptance Muon Spectrometer
• Coverage up to ? lt 2.7

Muons from ?, Z0 decays

• Si Tracker
• Large coverage up to ? lt 2.5
• Finely segmented pixel and strip detectors
• Good momentum resolution

Tracking particles with pT ? 1.0 GeV/c
2. 3.
Heavy quarks(b), quarkonium suppression(?, ?)
1. 3.
Global event characterization
7
Simulation Tools
Event Generator HIJING Based on PYTHIA and
Lund fragmentation scheme with nuclear effects
nuclear shadowing, jet quenching
Simulated event samples
HIJING full GEANT3 ATLAS detector
simulations Only particles within y lt 3.2

• High Geant cuts 1 MeV tracking/10 MeV production

• 5,000 events in each of 5 impact parameter bins
• b 0-1, 1-3, 3-6, 6-10, 10-15 fm

• Standard ATLAS cuts100 keV tracking/1 MeV
  production

• 1,000 central events, b 0-1fm

• Initial layout 2 pixel barrel layers

• 1,000 central events, b 0-1fm

8
Central PbPb Collision
Nch(y?0.5)

• About 75,000 stable particles
• 40,000 particles in ? ? 3
• CPU 6 h per central event (800MHz)
• Event size 50MB (without TRT)

9
Average Occupancies
Central Collision Events b0-1 fm

• Occupancies still reasonable in all Si
  Detectors
• below 2 in Pixels and below 20 in Strips
• (after accounting for local fluctuations in
  the data with low GEANT cuts)
• TRT unusable too high occupancy

10
Global Measurements
DAY-ONE MEASUREMENTS! Nch, dNch/d?, ?ET,
dET/d?, b

• Constrain model prediction
• Indispensable for all physics analyses

Predictions for PbPb central collisions at LHC
(dNch/d?)??0
Model/data
6500 HIJINGwith quenching, with
shadowing 3200 HIJINGno quenching,
with shadowing 2300 Saturation Model
(Kharzeev Nardi) 1500 Extrapolation
from lower energy data
11
Measurements of Nch(? lt 3)
Based on the correlation between measurable
quantity Q and the true number of charged
primary particles Q f(Nch)

Q Nsig (all Si detectors,except PixB-B)
?EtotEM, ?EtotHAD ?ETEM , ?ETHAD

• Caution
• Consistency between the measured signals
• and the simulated ones
• Monte Carlo dependency

12
Measurements of Nch(? lt 3)
Relative reconstruction errors Nrec-Nch/Nch
Reconstructed multiplicity distribution (Nsig)
Histogram true Nch
Points reconstructed Nch
Uncertainty up to 10 at low Nch, less than 3 at
high Nch
13
Reconstruction of dNch/d?
Motivation shape of the dNch/d? distribution is
sensitive to dynamical effects like
e.g. quenching and shadowing.

• Analysis is based on signals only from Pixel
  barrel layers
• (done separately for each layer).

• Clusterization procedure i.e. merging of hits in
  neighbor
• pixels is applied (particle traverses more
  than one pixel
• when ? ? 0).

• Correction factors need to be applied to account
  for the
• excess of clusters at large ?
• double hits from overlapping sensors
• magnetic field effects (low pT particles
  bending back)
• production of secondary particles

14
Reconstructed dNch/d?

• Comparison of the reconstructed dNch/d?
  distributions with the true one of charged
  primary particles.
• One single correction function C(?) calculated
  from
• a sample of central events is used.

Single PbPb event, b 0-1fm
5 peripheral collision, b 10-15fm
Reconstruction errors 5
Reconstruction errors 13
Correction factors are centrality independent!
15
Reconstructed dNch/d?
Single PbPb HIJING event with jet quenching, b
0-1fm
100 pp events at ?s200 GeV
Different shape and higher density are correctly
reproduced!!
Correction factors are insensitive to the
detailed properties of generated particles!
16
Estimate of the Collision Centrality
Based on the correlation between the measurable
quantity Q and the centrality parameter b,
(Npart, Ncoll)
Monotonic relation between Q and b allows for
assigning to a certain fraction of events
selected by cuts on Q, a well defined average
impact parameter.
Nsig
ET - EM
ET - HAD
17
Estimate of the Collision Centrality
Resolution of the estimated impact parameter
Remark A better approach would be to use a
quantity measured outside the
mid-rapidity region,e.g. energy in forward
calorimeters, which is less sensitive to
dynamical effects.
18
Track Reconstruction
Pixel and SCT detectors ATLAS xKalman
algorithm Starting with software release 6.1.0
all modifications specific to Heavy Ion track
reconstruction are included in the official
xKalman code.

• pT threshold for reconstructable
• tracks is 1 GeV (reduce CPU).
• Tracking cuts are optimized to
• get a decent efficiency and
• low rate of fake tracks.

• At least 10 measurements per track
• Maximum two shared measurements
• ?2/ndf ? 4

• Tracking in the ? lt 2.5
• For pT 1 - 15 GeV/c efficiency 70 fake
  ratelt10
• Fake rate at high pT can be reduced by matching
  with Calo data

19
Track Reconstruction
Momentum resolution
Efficiency versus rapidity
3 for pT up to 20 GeV/c (for ? lt 2.5)
Flat dependency for y lt 2 Higher in EC (more
layers)
Tracking in HI events looks promising, still can
be optimized!
20
Heavy-Quark Production

• Heavy quarks live through the thermalization of
  QGP
• ?can be affected by the presence of QGP
• Their radiative energy loss is qualitatively
  different
• than for light quarks.

Open Beauty via semi-leptonic decays

• Tagging of the B-jets
• the high pT ? in the MS
• displaced vertex in the ID

21
B-jet Tagging

• Preliminary study
• Standard ATLAS algorithm for pp
• Higgs events embedded into pp or Pb-Pb event
• Cuts on the vertex impact parameter in the Pixel
  and SCT

Rejection factors against light quarks versus
b-tagging efficiency
p-p
Pb-Pb
Promising, should be improved when combined with
muon tagging!
22
Quarkonium Suppression
Direct probe of the QGP Color screening of the
binding potential leads to the dissociation of
the quarkonium states.
Upsilon family
?(1s) ?(2s) ?(3s) Binding energies
(GeV) 1.1 0.54
0.2 Dissociation at the temperature 2.5Tc
0.9Tc 0.7Tc
Important to separate ?(1s) and ?(2s)
? ? ??
Upsilon mass reconstruction using the Muon
Spectrometer, Silicon Tracker and the Pixel
Detector (barrel sections only).
23
Quarkonium Suppression

• GEANT3 simulations of pure ?(1s) and ?(2s)
  states ? ??
• Muons with pT gt 3GeV are tracked backwards to
  the ID
• Invariant mass is calculated from the overall
  fit.

? 130 MeV

• Background estimate (HIJINGG3) ? S/B 0.6
• Acceptance 10-15 providing 100 efficient
  dimuon trigger
• Overlay with HIJING Event is under study!

24
Summary

• ATLAS detector will be capable of measuring
• many aspects of relatively low pT heavy-ion
• physics
• Lets see the detector performance for
• studying the truly high pT phenomena

25
BACKUPS
26
Detector Occupancies
Examples of occupancy versus z and Nch (high
GEANT thresholds)
Occ
Occ
z
z
Nch
Nch
Pix1
SCT1
27
Detector occupancies
Central collisions b0-1 fm, low GEANT thresholds
Pixel Detector
Silicon Tracker
28
Trigger DAQ
For PbPb collisions the interaction rate is
8kHz, a factor of 10 smaller than LVL 1
bandwidth. We expect further reduction to 1kHz by
requiring central collisions and pre-scaled
minimum bias events (or high pT jets or
muons). The event size for a central collision is
5 Mbytes. Similar bandwidth to storage as pp at
design L implies that we can afford 50 Hz data
recording.
200 Hz
29
Correction Factors
Correction factors are defined as
C(?) calculated from the sample of 50
central (b0-1fm) PbPb events, and then
parameterized.
Correction function for the inner most barrel
layer.
30
Maximal Cluster Size
Define the expected maximal size of the cluster

• In Z-direction the number of pixels to be merged
  
• depends on the Z-coordinate of the hit

e.g. for R5cm, Zhit40cm Npixel ? 6 -7

• In ?-direction the number of traversed pixels
  depends on pT.
• For a track with a curvature r, an angle at
  which particle
• enters the sensor is cos(?)R/2r (assuming
  that sensors form
• an ideal tube).
• Taking r 15cm (corresponding to pT90 MeV/c)

Npixels 4 6 for R12cm Npixels 3 4 for
R 5cm
31
Cluster Formation

• Choose seeds ? large signals gt 10,000 electrons
• Start with the seed with the largest signal
• Attached to it a signal in the adjacent pixel as
  long as

• There is a signal in a pixel
• One of the closest neighbor pixels already
  belongs to the cluster
• The distance from the seed to the pixel is not
  larger than the
• expected maximal size of the cluster (in both Z
  and ? directions)
• up to ?6 pixels in Z (depending on Zhit) and ?3
  pixels in ? (depending on R)