- Email: [email protected]
Z -jet correlations in lead–lead collisions at the LHC with ATLAS
Available online at www.sciencedirect.com
Nuclear Physics A 904–905 (2013) 233c–240c www.elsevier.com/locate/nuclphysa
Photon and Z production, and γ/Z-jet correlations in lead-lead collisions at the LHC with ATLAS Peter Steinberg (for the ATLAS Collaboration)1 Brookhaven National Laboratory, Upton, NY 11973 USA
Abstract ATLAS results on the production of photons and Z bosons in lead-lead collisions at the LHC are presented. Their production rates are found to scale with the number of binary collisions, as expected from QCD factorization at large transverse momentum. This makes them an ideal tool for probing the modification of jets which balance their transverse momentum. Despite the large differences in available statistics, both channels show a strong modification of the energy of the recoil jets, which increases with centrality.
1. Introduction Dijet events in heavy ion collisions were found to show strong asymmetries in the transverse energy of the two jets. However, these studies have been limited by the fact that it is not clear whether only one or both of the jets is modified by the medium [1]. Photons and Z bosons (and their decay leptons) are not affected by the hot, dense medium and thus should allow the calibration of the scale of the hard process, and thus directly probe energy loss. Furthermore, their individual rates are useful to check production rates which are calculable in pQCD, whether at NLO (for photons) or NNLO (for Z bosons). Of course photons have an appreciable rate to be emitted in the fragmentation of high energy jets, so an isolation condition (e.g. a maximum energy in a well-defined cone around the photon) is often applied to make the comparison to the calculations well-defined. In either case, both photons and Z’s are sensitive to modifications to the parton distribution functions in the nuclear environment. 2. Experimental setup and data taking The ATLAS detector [2] is a capable detector for measuring both photons and Z bosons in the high multiplicity heavy ion environment. The high resolution inner detector (ID), composed of silicon pixels, silicon strips, and a transition radiation tracker, covers up to |η| < 2.5 and provides a high resolution energy measurement for electrons and muons. The electromagnetic showers from photons and electrons are measured in a longitudinally-segmented electromagnetic calorimeter, which covers |η| < 4.9, with particularly fine segmentation within |η| < 2.5. Muons 1A
list of members of the ATLAS Collaboration and acknowledgements can be found at the end of this issue. © CERN for the benefit of the ATLAS Collaboration.
0375-9474/ © 2013 CERN Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.nuclphysa.2013.01.064
234c
dN ⁄ dm [GeV-1]
P. Steinberg / Nuclear Physics A 904–905 (2013) 233c–240c ATLAS Pb+Pb sNN = 2.76 TeV Data 2011 Lint = 0.15 nb-1 Z→ee Opposite sign: 772 Same sign: 42
100
Z→μμ 1223 14 Simulation
50
0
70
80
90 100 110 mee [GeV]
70
80
90 100 110 mμμ [GeV]
Figure 1: Mass spectrum of Z bosons reconstructed in the dielectron (left) and dimuon (right) channels (from [6]).
are measured in the inner detector as well as a high resolution muon spectrometer (MS), covering |η| < 2.7. The measurements were made using the 2011 lead-lead dataset, taken in November and December 2011, with an integrated luminosity of Lint = 0.16 nb−1 , with (0.14-0.15) nb−1 usable for physics analysis. Both the photon and Z analyses use the ATLAS forward calorimeter (FCal) to define the event centrality on an event-by-event basis, using the total transverse momentum ( ET ) of all calorimeter cells at the electromagnetic scale in the interval 3.1 < |η| < 4.9. The FCal ET distribution in 2011 was found to be identical to that measured in 2010 (after a constant 4.1% rescaling of the electromagnetic energy scale) which was calibrated to sample 98±2% of the total inelastic cross-section. The uncertainties in the geometric parameters (the number of participating nucleons Npart , the number of binary collisions Ncoll , and the mean nuclear thickness function T AA ) include cross-section and Glauber uncertainties [3]. The photon analysis [3] uses Lint = 0.13 nb−1 of the 2011 data sample. A detailed calibration of the integrated luminosity scale to the number of minimum bias events gave a total of 7.6 × 108 events sampled within 0-80% centrality, with an error of less than 1%. The data were prepared as a special selection of events triggered on an electromagnetic cluster (in a window size of Δη × Δφ = 0.2 × 0.1 or 0.1 × 0.2) with a transverse energy of 16 GeV at the trigger scale. Using minimum bias data, this trigger is found to be 100% efficient for photons with a transverse energy above 25 GeV. Contributions to the overall energy flow from the underlying event are removed from every event, using an identical algorithm to that used for the ATLAS jet analysis [4, 5]. The jet analysis reconstructs jets using the anti-kt algorithm with a radius parameter R. It is based on excluding regions around R = 0.2 calorimeter jets with ET > 25 GeV and track-jets with ET > 10 GeV, removing the modulation due to elliptic flow, and then updating the set of excluded regions in a second iteration step, whereupon the mean background and elliptic flow correction is recalculated. The Z boson analysis [6] has been performed in two different channels, via decays to dielectrons and dimuons. The dielectron channel was triggered using a similar trigger as for the photons, but with a slightly lower threshold of ET > 14 GeV. Also, as with the photons, there was no use of the ATLAS high level trigger system. In the offline analysis, the underlying event background, calculated during the jet reconstruction (discussed previously), was used to correct the electron energy reconstructed with unsubtracted cells. The dimuon channel required full use
P. Steinberg / Nuclear Physics A 904–905 (2013) 233c–240c
235c
Figure 2: (left) Rapidity distribution of Z bosons in the 0-80% centrality interval, (middle and right) transverse momentum distributions. In both cases, the measured spectra are compared with the distribution from PYTHIA, normalized to the NNLO cross-section and scaled by T AA (from [6]).
of the ATLAS multi-level trigger system. Muons reconstructed with pT > 4 GeV at Level 1 were used to seed a track reconstructed by the high level trigger system using precision muon spectrometer information, as well as the inner detector. The high level trigger system was also able to trigger on fully reconstructed events in the muon spectrometer, for muons with pT > 10 GeV, seeded on a coincidence of the ATLAS Zero Degree Calorimeters. The Z analysis used nearly the full sample of Lint = 0.15 nb−1 , and thus sampled over 8 × 108 events in the 0-80% centrality interval. 3. Z yields The mass spectra of dielectron and dimuon pairs from the triggered lead-lead events are shown in Fig. 1. Reconstructed electrons require a match of a track with an electromagnetic cluster, along with a set of shower shape cuts chosen to give good efficiency, while rejecting background from hadronic activity within jets. In the Z → μμ analysis, single muons are reconstructed with several quality levels. High quality muons are reconstructed in both the MS and ID with consistent angular measurements, as well as with a good match to the event vertex. At least one muon in each pair, matched to the trigger, is required to be of such quality. If the second muon in the pair has hit patterns in the MS and ID satisfying criteria of high reconstruction quality, the minimum pT threshold is set to 10 GeV for both muons. If the second muon fails this condition, both muons are required to satisfy pT > 20 GeV. After these selections, 772 Z boson candidates are reconstructed in the dielectron channel, while 1223 candidates (with 3% same-sign background) are reconstructed in the dimuon channel [6]. The efficiency corrected rapidity and transverse momentum distribution of Z bosons are shown in the left and right panels of Fig. 2, respectively. They are compared to distributions from PYTHIA, with the overall integral normalized to the NNLO cross-section [7], scaled by the mean nuclear thickness function T AA . The yield of Z bosons, integrated over transverse momentum and rapidity, and scaled by the number of binary collisions is shown in Fig. 3. It is observed that within the statistical
P. Steinberg / Nuclear Physics A 904–905 (2013) 233c–240c Z→ll 109 N |y|<2.5 〈Ncoll 〉 Nevents
236c
6
ATLAS Pb+Pb sNN = 2.76 TeV Data 2011 Lint = 0.15 nb-1
Z → ee Z → ll Z → μμ
All pZ T
4 pZ<10GeV T
2
1030GeV T
0
100
200
300
400 〈Npart 〉
Figure 3: Yield of Z bosons, for four transverse momentum intervals (pT < 10 GeV, 10 < pT < 30 GeV, pT > 30 GeV. and all pT ), divided by Ncoll as a function of Npart (from [6]).
and systematic uncertainties, the yield of Z’s at both low and high pT scales linearly with Ncoll , indicating no strong modification of the nuclear parton distribution functions with increasing centrality. 4. Photon yields Photon reconstruction in the ATLAS calorimeter [8, 9] is seeded by clusters with ET > 2.5 GeV. Clusters are found using a sliding window algorithm, themselves seeded on local maxima in the second sampling layer, which captures over 50% of the transverse energy of a typical photon. The high track density precludes the matching of tracks from conversions, produced in the beam pipe of detector far from the primary vertex, to a photon candidate. However, converted photons with large transverse momentum desposit energy in only a slightly larger region than unconverted photons, so calibrations were determined for the inclusive photon sample. The energy measurement is made using the full electromagnetic calorimeter. Note that the background subtraction is a small correction,of the order of a GeV even in the most central events, due to the small size of electromagnetic showers in the calorimeter. Photons are distinguished from the dijet background by a selection on a set of nine shower shape variables, similar to that used in pp analyses [8, 9] but optimized for heavy ions [3] (called the ”HI Tight” selection), and an isolation criterion based on requiring a maximum of 6 GeV transverse energy deposited in the calorimeter in a cone of R=0.3 around the photon candidate direction. The photon purity is determined by the ”double sideband” technique, which uses the ratio of non-isolated to isolated ”non-tight” photons (which are enriched with dijets) to extrapolate non-isolated tight photons into the signal region of tight, isolated photons. The photon efficiency is determined from simulations as a function of centrality, and is the product of three efficiencies: reconstruction, identification, and isolation. The last factor is the probability of a signal photon passing the isolation requirement, which is found to be about 85% in central events, due to the width of the isolation transverse energy distribution stemming purely from underlying event fluctuations. The efficiency corrected yields of prompt, isolated photons as a function of centrality are shown in the left panel of Fig. 4, compared with CMS data [10] (from a slightly wider pseudo-
237c
107
ATLAS Preliminary
0-10%(× 102)
Pb+Pb sNN = 2.76 TeV
10-20%(× 101)
Lint = 133 μb
-1
|η|<1.3
20-40%(× 100) 40-80% JETPHOX 1.3.0
105
PYTHIA MC11 CMS HI (|η|<1.44)
Data/JETPHOX
(1/Nevt)dNγ /dpT / 〈TAA 〉 [pb/GeV]
P. Steinberg / Nuclear Physics A 904–905 (2013) 233c–240c 8 0-10%(+6)
6 10-20%(+4)
CMS pp (|η|<1.44)
103 4 20-40%(+2)
10 2
10-1
10-3 0
40-80%
100
0 0
200 photon p [GeV] T
ATLAS Preliminary Pb+Pb sNN = 2.76 TeV Lint = 133 μb-1 |η|<1.3
100
200
300
photon p [GeV] T
Figure 4: (left) Efficiency corrected yields of prompt photons in |η| < 1.3 using HI tight cuts, isolation cone radius Riso = 0.3 and isolation energy of 6 GeV, divided by JETPHOX 1.3 predictions, which implement the same isolation cut, from [3]. Statistical errors are shown by the error bars. Systematic uncertainties on the photon yields are combined and shown by the yellow bands. The scale uncertainties from T AA are not shown. (right) Corrected yields divided by JETPHOX 1.3 predictions, which implement the same isolation selection, from [3].
rapidity interval) and perturbative NLO calculations from JETPHOX 1.3.0 [11], with the same isolation requirement applied at the parton level. The right panel shows the data divided by the NLO calculations, which is comparable to RAA but relative to theory rather than pp data (which are not available for the photon transverse energy range considered here). These results are con√ sistent with measurements from the PHENIX experiment in gold-gold collisions at sNN = 200 √ GeV [12]. and CMS in lead-lead collisions at sNN = 2.76 TeV [10], which both also observed yields to be consistent with a linear scaling with Ncoll . 5. Boson-jet correlations To quantify the modification of jet production in heavy ion collisions, three variables are used, following the notation established by the first results from CMS [14]: the momentum fraction jet xJγ = pT /pγT , with an equivalent definition for xJZ ; the acoplanarity ΔΦJγ ; and the survival probability RJγ , the probability of finding a matching jet for a selected photon or Z. 5.1. Z-jet correlations For Z-jet correlations, the jets are reconstructed and background subtracted using the standard iterative procedure described above, and a fake rejection is applied based on the presence of a track-jet or electromagnetic cluster close to the jet direction. It is observed that for Z transverse momentum above 50 GeV, the jet and Z are emitted primarily back to back. To account for the smearing of the jet momentum by detector effects, a bin-by-bin unfolding in jet pT is performed. jet Fig. 5 shows the corrected xJZ distribution for pT > 25 GeV, pZT > 60 GeV and xJZ > 25/60, for 0-20% central events (left) and 20-80% central (right). Although the number of counts is quite
238c
P. Steinberg / Nuclear Physics A 904–905 (2013) 233c–240c
p Z > 60 GeV T p jet T p jet T
T
0.9
〈p
jet
T
/ pZ 〉
Figure 5: Ratio of the jet (R=0.3) and photon transverse momenta, for Z transverse momentum greater than 60 GeV and jet transverse momentum greater than 25 GeV. Left panel is for 0-20% most central events while right panel is for 20-80% most central, from [13].
PYTHIA Data Data (All Centrality)
> 25 GeV / p Z > 25/60
ATLAS Preliminary Pb+Pb sNN = 2.76 TeV Data 2011 Lint = 0.15 nb-1
T
0.8 0.7 0.6
Anti-k t Jet R=0.2
100
200
Anti-k t Jet R=0.3
300
〈Npart 〉
100
200
Anti-k t Jet R=0.4
300
〈Npart 〉
100
200
300
〈Npart 〉
Figure 6: Z-jet transverse momentum fraction as a function of Npart for three jet radius parameters. Two exclusive centrality selections (0-20% and 20-80%) and an inclusive selection (0-80%) are shown, from [13].
small, Fig. 6 shows that a systematic decrease in the mean xJZ is observed going from peripheral to central collisions, and the central value is significantly lower than expected from PYTHIA predictions. 5.2. Photon-jet correlations The advantage to photon-jet correlations are the large increase in cross-section, since the Z mass no longer constrains the available phase space. The disadvantage is, as mentioned above, the larger background from dijets. The photon-jet analysis from ATLAS is based on the yield analysis described above, although the primary simulations are performed by overlaying PYTHIA photon-jet events onto real data evets. These simulations are used to determine both photon and jet efficiencies, for events with the leading jet (after fake rejection) falling within Δφ > 7π/8 of a tight, isolated photon candidate. The use of data overlay removes uncertainties related to whether or not the background is correctly modeled, since it is so by construction. The response of the jets is unfolded for jet energy resolution using response matrices determined by the PYTHIA+Data samples. The unfolding is implemented in two steps using the SVD approach [16]. This is first performed on the jets from the inclusive sample of photon-jet candidate
239c
Pb+Pb sNN=2.76 TeV
0.5
0
ATLAS Preliminary
0.5
1
1.5
0.5
0
ATLAS Preliminary
0.5
1
1.5
1
Jγ
PYTHIA+Data Data 20-40% R=0.3 Pb+Pb sNN=2.76 TeV
1.5
1
Lint =0.13 nb-1 0.5
0 0.5
ATLAS Preliminary
1
1.5
xJγ
ATLAS Preliminary
0.5
1
0.5
ATLAS Preliminary
1
1.5
1.5
xJγ
PYTHIA+Data Data 0-10% R=0.2
1
Pb+Pb sNN=2.76 TeV Lint =0.13 nb-1
0.5
0
2
2 PYTHIA+Data Data 10-20% R=0.3 Pb+Pb sNN=2.76 TeV
1.5
1
ATLAS Preliminary
0.5
1
1.5
2
xJγ 2 PYTHIA+Data Data 0-10% R=0.3 Pb+Pb sNN=2.76 TeV
1.5
1
Lint =0.13 nb-1 0.5
0 0.5
2
1.5
xJγ
Lint =0.13 nb-1
0 0.5
2
0.5
0
2
2
Lint =0.13 nb-1
xJγ (1/Nγ ) dN /dx Jγ
(1/Nγ ) dNJγ /dx Jγ
PYTHIA+Data Data 40-80% R=0.3 Pb+Pb sNN=2.76 TeV
Pb+Pb sNN=2.76 TeV
2
xJγ
1.5
1
Lint =0.13 nb-1
2
2
1.5
PYTHIA+Data Data 10-20% R=0.2
(1/Nγ ) dNJγ /dx Jγ
1
Lint =0.13 nb-1
2
(1/Nγ ) dNJγ /dx Jγ
Pb+Pb sNN=2.76 TeV
1.5
PYTHIA+Data Data 20-40% R=0.2
(1/Nγ ) dNJγ /dx Jγ
1
2
(1/Nγ ) dNJγ /dx Jγ
1.5
PYTHIA+Data Data 40-80% R=0.2
Jγ
2
(1/Nγ ) dN /dx Jγ
(1/Nγ ) dNJγ /dx Jγ
P. Steinberg / Nuclear Physics A 904–905 (2013) 233c–240c
ATLAS Preliminary
1
1.5
Lint =0.13 nb-1 0.5
0 0.5
2
xJγ
ATLAS Preliminary
1
1.5
2
xJγ
Figure 7: Unfolded distribution of photon-jet transverse momentum fraction, normalized per photon in four centrality intervals, for photon transverse momentum between 60-90 GeV and jet momentum greater than 25 GeV. The top row is for jet radius R = 0.2 and the bottom row is for R = 0.3, from [15].
events, and then the unfolding matrix is applied event-by-event to map the pT of each leading jet into the final histogram of xJγ . Figure 7 shows the first fully unfolded results for (1/Nγ )dNJγ /dxJγ in lead-lead collisions as a function of centrality (four bins: 40-80%, 20-40%, 10-20% and 0-10%) and for two jet radius parameters (R = 0.2 and R = 0.3). The photon is selected to have 60 < pγT < 90 GeV and jet |ηγ | < 1.3 The associated leading jet is selected to have pT > 25 GeV, |ηjet | < 2.1 and |ΔφJγ | > γ 7π/8. To compensate for the large photon pT interval, events are only kept if xJγ > 25/60. The error bars are statistical, although the unfolding procedure induces correlations between nearby bins. The grey error bands include the total systematic uncertainties with contributions from both the jet and photon. The yellow histogram shows PYTHIA results using true photon and true jet kinematics in events selected with the same isolation and photon selection criteria as that used in reconstructed events. It is observed for both jet radius parameters that while the xJγ distribution measured in peripheral events is similar to that predicted by PYTHIA, the central events show a strongly modified distribution and a strong reduction in the overall yield, from the reduction of the number of events satisfying the minimum xJγ requirement. These modifications are quantified more precisely in the left and right panels of Fig. 8. The left two figures show xJγ as a function of Npart for jet radius parameters R = 0.2 and R = 0.3. It is observed that while the measured mean agrees with PYTHIA predictions in the most peripheral events, it is significantly reduced in the most central events. The right two figures show the Npart evolution of RJγ , the fraction of photon events with an associated jet satisfying the kinematic requirements. While peripheral events agree with the PYTHIA prediction that 70% of photons have associated jets in an unmodified distribution, this fraction is reduced by nearly a factor of two in the most central events. Despite these strong modifications to the jet energy in central events, exponential fits to the acoplanarity relative to the photon show no significant change to the slope as a function of centrality [15].
0.5
1
0.5 R=0.2 Data
1
0.5 R=0.3 Data
1
0.5 R=0.2 Data
R=0.3 Data
R=0.2 PYTHIA+Data
R=0.3 PYTHIA+Data
R=0.2 PYTHIA+Data
R=0.3 PYTHIA+Data
ATLAS Preliminary
ATLAS Preliminary
ATLAS Preliminary
ATLAS Preliminary
Pb+Pb Lint =0.13 nb-1
Pb+Pb Lint =0.13 nb-1
Pb+Pb Lint =0.13 nb-1
Pb+Pb Lint =0.13 nb-1
sNN=2.76 TeV 0 0
RJγ
1
RJγ
P. Steinberg / Nuclear Physics A 904–905 (2013) 233c–240c 〈x Jγ 〉
〈x Jγ 〉
240c
100
200
sNN=2.76 TeV 300
400
0 0
Npart
100
200
sNN=2.76 TeV 300
400
Npart
0 0
100
200
sNN=2.76 TeV 300
400
0 0
100
200
300
Npart
400
Npart
Figure 8: (left) Mean photon-jet momentum fraction as a function of Npart for two jet radius parameters (0.2 and 0.3) (right) Mean photon-jet survival fraction, also as a function of Npart for two jet radius parameters (0.2 and 0.3), from Ref [15].
6. Conclusion and outlook Electroweak probes provide a powerful tool to study the modification of jets in the hot, dense medium produced in heavy ion collisions. This work has shown recent ATLAS results on Z and photon production, and new results on their correlation with jets. Measurements of Z bosons show that the rapidity and momentum distributions agree well with PYTHIA calculations normalized to the NNLO cross-section and scaled by T AA . The integrated yields scale linearly with the number of binary collisions. Finally, the correlation of Z’s with jets show a clear attenuation of the jet energy relative to the expectation from the Z pT . Photon yields are measured out to 200 GeV, after accounting for the background from dijet events. Good agreement with JETPHOX 1.3.0 is observed, along with linear scaling with Ncoll . The correlation of photons with jets, after full corrections and unfolding, shows a clear back-to-back correlation but a strong centrality-dependent attenuation of the jet energy and a reduction of the fraction of photons associated with a jet. While these measurements already provide a first look at jet quenching without the ambiguities inherent in dijet correlations, they will both clearly benefit from the increased energy and luminosity of the LHC after the upcoming long shutdown. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16]
ATLAS Collaboration, Phys. Rev. Lett. 105, 252303 (2010). ATLAS Collaboration, JINST 3, S08003 (2008). ATLAS Collaboration, ATLAS-CONF-2012-051 (2012). http://cdsweb.cern.ch/record/1451913. ATLAS Collaboration, J. Phys. G 38, 124021 (2011). ATLAS Collaboration, arXiv:1208.1967 (2012). ATLAS Collaboration, arXiv:1210.6486 (2012). ATLAS Collaboration, JHEP 1012 060 (2010). ATLAS Collaboration, Phys. Rev. D 83, 052005 (2011). ATLAS Collaboration, Phys. Lett. B 706, 150 (2011). S. Chatrchyan et al. [CMS Collaboration], Phys. Lett. B 710, 256 (2012). P. Aurenche, M. Fontannaz, J. -P. Guillet, E. Pilon and M. Werlen, Phys. Rev. D 73, 094007 (2006). S. S. Adler et al. [PHENIX Collaboration], Phys. Rev. Lett. 94, 232301 (2005). ATLAS Collaboration, ATLAS-CONF-2012-119 (2012). http://cdsweb.cern.ch/record/1472941. S. Chatrchyan et al. [CMS Collaboration], arXiv:1205.0206 [nucl-ex]. ATLAS Collaboration, ATLAS-CONF-2012-121 (2012). http://cdsweb.cern.ch/record/1473135. A. Hocker and V. Kartvelishvili, Nucl. Instrum. Meth. A372, 469 (1996).
Nuclear Physics A 904–905 (2013) 233c–240c www.elsevier.com/locate/nuclphysa
Photon and Z production, and γ/Z-jet correlations in lead-lead collisions at the LHC with ATLAS Peter Steinberg (for the ATLAS Collaboration)1 Brookhaven National Laboratory, Upton, NY 11973 USA
Abstract ATLAS results on the production of photons and Z bosons in lead-lead collisions at the LHC are presented. Their production rates are found to scale with the number of binary collisions, as expected from QCD factorization at large transverse momentum. This makes them an ideal tool for probing the modification of jets which balance their transverse momentum. Despite the large differences in available statistics, both channels show a strong modification of the energy of the recoil jets, which increases with centrality.
1. Introduction Dijet events in heavy ion collisions were found to show strong asymmetries in the transverse energy of the two jets. However, these studies have been limited by the fact that it is not clear whether only one or both of the jets is modified by the medium [1]. Photons and Z bosons (and their decay leptons) are not affected by the hot, dense medium and thus should allow the calibration of the scale of the hard process, and thus directly probe energy loss. Furthermore, their individual rates are useful to check production rates which are calculable in pQCD, whether at NLO (for photons) or NNLO (for Z bosons). Of course photons have an appreciable rate to be emitted in the fragmentation of high energy jets, so an isolation condition (e.g. a maximum energy in a well-defined cone around the photon) is often applied to make the comparison to the calculations well-defined. In either case, both photons and Z’s are sensitive to modifications to the parton distribution functions in the nuclear environment. 2. Experimental setup and data taking The ATLAS detector [2] is a capable detector for measuring both photons and Z bosons in the high multiplicity heavy ion environment. The high resolution inner detector (ID), composed of silicon pixels, silicon strips, and a transition radiation tracker, covers up to |η| < 2.5 and provides a high resolution energy measurement for electrons and muons. The electromagnetic showers from photons and electrons are measured in a longitudinally-segmented electromagnetic calorimeter, which covers |η| < 4.9, with particularly fine segmentation within |η| < 2.5. Muons 1A
list of members of the ATLAS Collaboration and acknowledgements can be found at the end of this issue. © CERN for the benefit of the ATLAS Collaboration.
0375-9474/ © 2013 CERN Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.nuclphysa.2013.01.064
234c
dN ⁄ dm [GeV-1]
P. Steinberg / Nuclear Physics A 904–905 (2013) 233c–240c ATLAS Pb+Pb sNN = 2.76 TeV Data 2011 Lint = 0.15 nb-1 Z→ee Opposite sign: 772 Same sign: 42
100
Z→μμ 1223 14 Simulation
50
0
70
80
90 100 110 mee [GeV]
70
80
90 100 110 mμμ [GeV]
Figure 1: Mass spectrum of Z bosons reconstructed in the dielectron (left) and dimuon (right) channels (from [6]).
are measured in the inner detector as well as a high resolution muon spectrometer (MS), covering |η| < 2.7. The measurements were made using the 2011 lead-lead dataset, taken in November and December 2011, with an integrated luminosity of Lint = 0.16 nb−1 , with (0.14-0.15) nb−1 usable for physics analysis. Both the photon and Z analyses use the ATLAS forward calorimeter (FCal) to define the event centrality on an event-by-event basis, using the total transverse momentum ( ET ) of all calorimeter cells at the electromagnetic scale in the interval 3.1 < |η| < 4.9. The FCal ET distribution in 2011 was found to be identical to that measured in 2010 (after a constant 4.1% rescaling of the electromagnetic energy scale) which was calibrated to sample 98±2% of the total inelastic cross-section. The uncertainties in the geometric parameters (the number of participating nucleons Npart , the number of binary collisions Ncoll , and the mean nuclear thickness function T AA ) include cross-section and Glauber uncertainties [3]. The photon analysis [3] uses Lint = 0.13 nb−1 of the 2011 data sample. A detailed calibration of the integrated luminosity scale to the number of minimum bias events gave a total of 7.6 × 108 events sampled within 0-80% centrality, with an error of less than 1%. The data were prepared as a special selection of events triggered on an electromagnetic cluster (in a window size of Δη × Δφ = 0.2 × 0.1 or 0.1 × 0.2) with a transverse energy of 16 GeV at the trigger scale. Using minimum bias data, this trigger is found to be 100% efficient for photons with a transverse energy above 25 GeV. Contributions to the overall energy flow from the underlying event are removed from every event, using an identical algorithm to that used for the ATLAS jet analysis [4, 5]. The jet analysis reconstructs jets using the anti-kt algorithm with a radius parameter R. It is based on excluding regions around R = 0.2 calorimeter jets with ET > 25 GeV and track-jets with ET > 10 GeV, removing the modulation due to elliptic flow, and then updating the set of excluded regions in a second iteration step, whereupon the mean background and elliptic flow correction is recalculated. The Z boson analysis [6] has been performed in two different channels, via decays to dielectrons and dimuons. The dielectron channel was triggered using a similar trigger as for the photons, but with a slightly lower threshold of ET > 14 GeV. Also, as with the photons, there was no use of the ATLAS high level trigger system. In the offline analysis, the underlying event background, calculated during the jet reconstruction (discussed previously), was used to correct the electron energy reconstructed with unsubtracted cells. The dimuon channel required full use
P. Steinberg / Nuclear Physics A 904–905 (2013) 233c–240c
235c
Figure 2: (left) Rapidity distribution of Z bosons in the 0-80% centrality interval, (middle and right) transverse momentum distributions. In both cases, the measured spectra are compared with the distribution from PYTHIA, normalized to the NNLO cross-section and scaled by T AA (from [6]).
of the ATLAS multi-level trigger system. Muons reconstructed with pT > 4 GeV at Level 1 were used to seed a track reconstructed by the high level trigger system using precision muon spectrometer information, as well as the inner detector. The high level trigger system was also able to trigger on fully reconstructed events in the muon spectrometer, for muons with pT > 10 GeV, seeded on a coincidence of the ATLAS Zero Degree Calorimeters. The Z analysis used nearly the full sample of Lint = 0.15 nb−1 , and thus sampled over 8 × 108 events in the 0-80% centrality interval. 3. Z yields The mass spectra of dielectron and dimuon pairs from the triggered lead-lead events are shown in Fig. 1. Reconstructed electrons require a match of a track with an electromagnetic cluster, along with a set of shower shape cuts chosen to give good efficiency, while rejecting background from hadronic activity within jets. In the Z → μμ analysis, single muons are reconstructed with several quality levels. High quality muons are reconstructed in both the MS and ID with consistent angular measurements, as well as with a good match to the event vertex. At least one muon in each pair, matched to the trigger, is required to be of such quality. If the second muon in the pair has hit patterns in the MS and ID satisfying criteria of high reconstruction quality, the minimum pT threshold is set to 10 GeV for both muons. If the second muon fails this condition, both muons are required to satisfy pT > 20 GeV. After these selections, 772 Z boson candidates are reconstructed in the dielectron channel, while 1223 candidates (with 3% same-sign background) are reconstructed in the dimuon channel [6]. The efficiency corrected rapidity and transverse momentum distribution of Z bosons are shown in the left and right panels of Fig. 2, respectively. They are compared to distributions from PYTHIA, with the overall integral normalized to the NNLO cross-section [7], scaled by the mean nuclear thickness function T AA . The yield of Z bosons, integrated over transverse momentum and rapidity, and scaled by the number of binary collisions is shown in Fig. 3. It is observed that within the statistical
P. Steinberg / Nuclear Physics A 904–905 (2013) 233c–240c Z→ll 109 N |y|<2.5 〈Ncoll 〉 Nevents
236c
6
ATLAS Pb+Pb sNN = 2.76 TeV Data 2011 Lint = 0.15 nb-1
Z → ee Z → ll Z → μμ
All pZ T
4 pZ<10GeV T
2
10
0
100
200
300
400 〈Npart 〉
Figure 3: Yield of Z bosons, for four transverse momentum intervals (pT < 10 GeV, 10 < pT < 30 GeV, pT > 30 GeV. and all pT ), divided by Ncoll as a function of Npart (from [6]).
and systematic uncertainties, the yield of Z’s at both low and high pT scales linearly with Ncoll , indicating no strong modification of the nuclear parton distribution functions with increasing centrality. 4. Photon yields Photon reconstruction in the ATLAS calorimeter [8, 9] is seeded by clusters with ET > 2.5 GeV. Clusters are found using a sliding window algorithm, themselves seeded on local maxima in the second sampling layer, which captures over 50% of the transverse energy of a typical photon. The high track density precludes the matching of tracks from conversions, produced in the beam pipe of detector far from the primary vertex, to a photon candidate. However, converted photons with large transverse momentum desposit energy in only a slightly larger region than unconverted photons, so calibrations were determined for the inclusive photon sample. The energy measurement is made using the full electromagnetic calorimeter. Note that the background subtraction is a small correction,of the order of a GeV even in the most central events, due to the small size of electromagnetic showers in the calorimeter. Photons are distinguished from the dijet background by a selection on a set of nine shower shape variables, similar to that used in pp analyses [8, 9] but optimized for heavy ions [3] (called the ”HI Tight” selection), and an isolation criterion based on requiring a maximum of 6 GeV transverse energy deposited in the calorimeter in a cone of R=0.3 around the photon candidate direction. The photon purity is determined by the ”double sideband” technique, which uses the ratio of non-isolated to isolated ”non-tight” photons (which are enriched with dijets) to extrapolate non-isolated tight photons into the signal region of tight, isolated photons. The photon efficiency is determined from simulations as a function of centrality, and is the product of three efficiencies: reconstruction, identification, and isolation. The last factor is the probability of a signal photon passing the isolation requirement, which is found to be about 85% in central events, due to the width of the isolation transverse energy distribution stemming purely from underlying event fluctuations. The efficiency corrected yields of prompt, isolated photons as a function of centrality are shown in the left panel of Fig. 4, compared with CMS data [10] (from a slightly wider pseudo-
237c
107
ATLAS Preliminary
0-10%(× 102)
Pb+Pb sNN = 2.76 TeV
10-20%(× 101)
Lint = 133 μb
-1
|η|<1.3
20-40%(× 100) 40-80% JETPHOX 1.3.0
105
PYTHIA MC11 CMS HI (|η|<1.44)
Data/JETPHOX
(1/Nevt)dNγ /dpT / 〈TAA 〉 [pb/GeV]
P. Steinberg / Nuclear Physics A 904–905 (2013) 233c–240c 8 0-10%(+6)
6 10-20%(+4)
CMS pp (|η|<1.44)
103 4 20-40%(+2)
10 2
10-1
10-3 0
40-80%
100
0 0
200 photon p [GeV] T
ATLAS Preliminary Pb+Pb sNN = 2.76 TeV Lint = 133 μb-1 |η|<1.3
100
200
300
photon p [GeV] T
Figure 4: (left) Efficiency corrected yields of prompt photons in |η| < 1.3 using HI tight cuts, isolation cone radius Riso = 0.3 and isolation energy of 6 GeV, divided by JETPHOX 1.3 predictions, which implement the same isolation cut, from [3]. Statistical errors are shown by the error bars. Systematic uncertainties on the photon yields are combined and shown by the yellow bands. The scale uncertainties from T AA are not shown. (right) Corrected yields divided by JETPHOX 1.3 predictions, which implement the same isolation selection, from [3].
rapidity interval) and perturbative NLO calculations from JETPHOX 1.3.0 [11], with the same isolation requirement applied at the parton level. The right panel shows the data divided by the NLO calculations, which is comparable to RAA but relative to theory rather than pp data (which are not available for the photon transverse energy range considered here). These results are con√ sistent with measurements from the PHENIX experiment in gold-gold collisions at sNN = 200 √ GeV [12]. and CMS in lead-lead collisions at sNN = 2.76 TeV [10], which both also observed yields to be consistent with a linear scaling with Ncoll . 5. Boson-jet correlations To quantify the modification of jet production in heavy ion collisions, three variables are used, following the notation established by the first results from CMS [14]: the momentum fraction jet xJγ = pT /pγT , with an equivalent definition for xJZ ; the acoplanarity ΔΦJγ ; and the survival probability RJγ , the probability of finding a matching jet for a selected photon or Z. 5.1. Z-jet correlations For Z-jet correlations, the jets are reconstructed and background subtracted using the standard iterative procedure described above, and a fake rejection is applied based on the presence of a track-jet or electromagnetic cluster close to the jet direction. It is observed that for Z transverse momentum above 50 GeV, the jet and Z are emitted primarily back to back. To account for the smearing of the jet momentum by detector effects, a bin-by-bin unfolding in jet pT is performed. jet Fig. 5 shows the corrected xJZ distribution for pT > 25 GeV, pZT > 60 GeV and xJZ > 25/60, for 0-20% central events (left) and 20-80% central (right). Although the number of counts is quite
238c
P. Steinberg / Nuclear Physics A 904–905 (2013) 233c–240c
p Z > 60 GeV T p jet T p jet T
T
0.9
〈p
jet
T
/ pZ 〉
Figure 5: Ratio of the jet (R=0.3) and photon transverse momenta, for Z transverse momentum greater than 60 GeV and jet transverse momentum greater than 25 GeV. Left panel is for 0-20% most central events while right panel is for 20-80% most central, from [13].
PYTHIA Data Data (All Centrality)
> 25 GeV / p Z > 25/60
ATLAS Preliminary Pb+Pb sNN = 2.76 TeV Data 2011 Lint = 0.15 nb-1
T
0.8 0.7 0.6
Anti-k t Jet R=0.2
100
200
Anti-k t Jet R=0.3
300
〈Npart 〉
100
200
Anti-k t Jet R=0.4
300
〈Npart 〉
100
200
300
〈Npart 〉
Figure 6: Z-jet transverse momentum fraction as a function of Npart for three jet radius parameters. Two exclusive centrality selections (0-20% and 20-80%) and an inclusive selection (0-80%) are shown, from [13].
small, Fig. 6 shows that a systematic decrease in the mean xJZ is observed going from peripheral to central collisions, and the central value is significantly lower than expected from PYTHIA predictions. 5.2. Photon-jet correlations The advantage to photon-jet correlations are the large increase in cross-section, since the Z mass no longer constrains the available phase space. The disadvantage is, as mentioned above, the larger background from dijets. The photon-jet analysis from ATLAS is based on the yield analysis described above, although the primary simulations are performed by overlaying PYTHIA photon-jet events onto real data evets. These simulations are used to determine both photon and jet efficiencies, for events with the leading jet (after fake rejection) falling within Δφ > 7π/8 of a tight, isolated photon candidate. The use of data overlay removes uncertainties related to whether or not the background is correctly modeled, since it is so by construction. The response of the jets is unfolded for jet energy resolution using response matrices determined by the PYTHIA+Data samples. The unfolding is implemented in two steps using the SVD approach [16]. This is first performed on the jets from the inclusive sample of photon-jet candidate
239c
Pb+Pb sNN=2.76 TeV
0.5
0
ATLAS Preliminary
0.5
1
1.5
0.5
0
ATLAS Preliminary
0.5
1
1.5
1
Jγ
PYTHIA+Data Data 20-40% R=0.3 Pb+Pb sNN=2.76 TeV
1.5
1
Lint =0.13 nb-1 0.5
0 0.5
ATLAS Preliminary
1
1.5
xJγ
ATLAS Preliminary
0.5
1
0.5
ATLAS Preliminary
1
1.5
1.5
xJγ
PYTHIA+Data Data 0-10% R=0.2
1
Pb+Pb sNN=2.76 TeV Lint =0.13 nb-1
0.5
0
2
2 PYTHIA+Data Data 10-20% R=0.3 Pb+Pb sNN=2.76 TeV
1.5
1
ATLAS Preliminary
0.5
1
1.5
2
xJγ 2 PYTHIA+Data Data 0-10% R=0.3 Pb+Pb sNN=2.76 TeV
1.5
1
Lint =0.13 nb-1 0.5
0 0.5
2
1.5
xJγ
Lint =0.13 nb-1
0 0.5
2
0.5
0
2
2
Lint =0.13 nb-1
xJγ (1/Nγ ) dN /dx Jγ
(1/Nγ ) dNJγ /dx Jγ
PYTHIA+Data Data 40-80% R=0.3 Pb+Pb sNN=2.76 TeV
Pb+Pb sNN=2.76 TeV
2
xJγ
1.5
1
Lint =0.13 nb-1
2
2
1.5
PYTHIA+Data Data 10-20% R=0.2
(1/Nγ ) dNJγ /dx Jγ
1
Lint =0.13 nb-1
2
(1/Nγ ) dNJγ /dx Jγ
Pb+Pb sNN=2.76 TeV
1.5
PYTHIA+Data Data 20-40% R=0.2
(1/Nγ ) dNJγ /dx Jγ
1
2
(1/Nγ ) dNJγ /dx Jγ
1.5
PYTHIA+Data Data 40-80% R=0.2
Jγ
2
(1/Nγ ) dN /dx Jγ
(1/Nγ ) dNJγ /dx Jγ
P. Steinberg / Nuclear Physics A 904–905 (2013) 233c–240c
ATLAS Preliminary
1
1.5
Lint =0.13 nb-1 0.5
0 0.5
2
xJγ
ATLAS Preliminary
1
1.5
2
xJγ
Figure 7: Unfolded distribution of photon-jet transverse momentum fraction, normalized per photon in four centrality intervals, for photon transverse momentum between 60-90 GeV and jet momentum greater than 25 GeV. The top row is for jet radius R = 0.2 and the bottom row is for R = 0.3, from [15].
events, and then the unfolding matrix is applied event-by-event to map the pT of each leading jet into the final histogram of xJγ . Figure 7 shows the first fully unfolded results for (1/Nγ )dNJγ /dxJγ in lead-lead collisions as a function of centrality (four bins: 40-80%, 20-40%, 10-20% and 0-10%) and for two jet radius parameters (R = 0.2 and R = 0.3). The photon is selected to have 60 < pγT < 90 GeV and jet |ηγ | < 1.3 The associated leading jet is selected to have pT > 25 GeV, |ηjet | < 2.1 and |ΔφJγ | > γ 7π/8. To compensate for the large photon pT interval, events are only kept if xJγ > 25/60. The error bars are statistical, although the unfolding procedure induces correlations between nearby bins. The grey error bands include the total systematic uncertainties with contributions from both the jet and photon. The yellow histogram shows PYTHIA results using true photon and true jet kinematics in events selected with the same isolation and photon selection criteria as that used in reconstructed events. It is observed for both jet radius parameters that while the xJγ distribution measured in peripheral events is similar to that predicted by PYTHIA, the central events show a strongly modified distribution and a strong reduction in the overall yield, from the reduction of the number of events satisfying the minimum xJγ requirement. These modifications are quantified more precisely in the left and right panels of Fig. 8. The left two figures show xJγ as a function of Npart for jet radius parameters R = 0.2 and R = 0.3. It is observed that while the measured mean agrees with PYTHIA predictions in the most peripheral events, it is significantly reduced in the most central events. The right two figures show the Npart evolution of RJγ , the fraction of photon events with an associated jet satisfying the kinematic requirements. While peripheral events agree with the PYTHIA prediction that 70% of photons have associated jets in an unmodified distribution, this fraction is reduced by nearly a factor of two in the most central events. Despite these strong modifications to the jet energy in central events, exponential fits to the acoplanarity relative to the photon show no significant change to the slope as a function of centrality [15].
0.5
1
0.5 R=0.2 Data
1
0.5 R=0.3 Data
1
0.5 R=0.2 Data
R=0.3 Data
R=0.2 PYTHIA+Data
R=0.3 PYTHIA+Data
R=0.2 PYTHIA+Data
R=0.3 PYTHIA+Data
ATLAS Preliminary
ATLAS Preliminary
ATLAS Preliminary
ATLAS Preliminary
Pb+Pb Lint =0.13 nb-1
Pb+Pb Lint =0.13 nb-1
Pb+Pb Lint =0.13 nb-1
Pb+Pb Lint =0.13 nb-1
sNN=2.76 TeV 0 0
RJγ
1
RJγ
P. Steinberg / Nuclear Physics A 904–905 (2013) 233c–240c 〈x Jγ 〉
〈x Jγ 〉
240c
100
200
sNN=2.76 TeV 300
400
0 0
Npart
100
200
sNN=2.76 TeV 300
400
Npart
0 0
100
200
sNN=2.76 TeV 300
400
0 0
100
200
300
Npart
400
Npart
Figure 8: (left) Mean photon-jet momentum fraction as a function of Npart for two jet radius parameters (0.2 and 0.3) (right) Mean photon-jet survival fraction, also as a function of Npart for two jet radius parameters (0.2 and 0.3), from Ref [15].
6. Conclusion and outlook Electroweak probes provide a powerful tool to study the modification of jets in the hot, dense medium produced in heavy ion collisions. This work has shown recent ATLAS results on Z and photon production, and new results on their correlation with jets. Measurements of Z bosons show that the rapidity and momentum distributions agree well with PYTHIA calculations normalized to the NNLO cross-section and scaled by T AA . The integrated yields scale linearly with the number of binary collisions. Finally, the correlation of Z’s with jets show a clear attenuation of the jet energy relative to the expectation from the Z pT . Photon yields are measured out to 200 GeV, after accounting for the background from dijet events. Good agreement with JETPHOX 1.3.0 is observed, along with linear scaling with Ncoll . The correlation of photons with jets, after full corrections and unfolding, shows a clear back-to-back correlation but a strong centrality-dependent attenuation of the jet energy and a reduction of the fraction of photons associated with a jet. While these measurements already provide a first look at jet quenching without the ambiguities inherent in dijet correlations, they will both clearly benefit from the increased energy and luminosity of the LHC after the upcoming long shutdown. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16]
ATLAS Collaboration, Phys. Rev. Lett. 105, 252303 (2010). ATLAS Collaboration, JINST 3, S08003 (2008). ATLAS Collaboration, ATLAS-CONF-2012-051 (2012). http://cdsweb.cern.ch/record/1451913. ATLAS Collaboration, J. Phys. G 38, 124021 (2011). ATLAS Collaboration, arXiv:1208.1967 (2012). ATLAS Collaboration, arXiv:1210.6486 (2012). ATLAS Collaboration, JHEP 1012 060 (2010). ATLAS Collaboration, Phys. Rev. D 83, 052005 (2011). ATLAS Collaboration, Phys. Lett. B 706, 150 (2011). S. Chatrchyan et al. [CMS Collaboration], Phys. Lett. B 710, 256 (2012). P. Aurenche, M. Fontannaz, J. -P. Guillet, E. Pilon and M. Werlen, Phys. Rev. D 73, 094007 (2006). S. S. Adler et al. [PHENIX Collaboration], Phys. Rev. Lett. 94, 232301 (2005). ATLAS Collaboration, ATLAS-CONF-2012-119 (2012). http://cdsweb.cern.ch/record/1472941. S. Chatrchyan et al. [CMS Collaboration], arXiv:1205.0206 [nucl-ex]. ATLAS Collaboration, ATLAS-CONF-2012-121 (2012). http://cdsweb.cern.ch/record/1473135. A. Hocker and V. Kartvelishvili, Nucl. Instrum. Meth. A372, 469 (1996).