EXCLUSIVE γγ → ℓ+ℓ- AND γp → Υp → ℓ+ℓ-p PRODUCTION AT CMS

Exclusive dilepton events are characterized by the presence of two back-to-back leptons, and no other detector activity above threshold. In the CMS experiment, this signature can result from two-photon interactions ( γγ → `+`−) or Υ photoproduction ( γp→ Υp→ `+`−p). Presented at Lake Louise Winter Institute,February 17-23, 2008,Lake Louise,Canada May 21, 2008 13:26 Proceedings Trim Size: 9in x 6in cmsdileptons2 EXCLUSIVE γγ → ll AND γp → Υp → llp PRODUCTION AT CMS J. HOLLAR, ON BEHALF OF THE CMS COLLABORATION Lawrence Livermore National Laboratory, 7000 East Avenue, Livermore, CA 94550, USA E-mail: jjhollar@llnl.gov Exclusive dilepton events are characterized by the presence of two back-to-back leptons, and no other detector activity above threshold. In the CMS experiment, this signature can result from two-photon interactions (γγ → ll) or Υ photoproduction (γp → Υp → llp).


Introduction
At the Large Hadron Collider, dilepton production through the processes γγ → ℓ + ℓ − and γp → Υp → ℓ + ℓ − p (Figure 1) is characterized by the presence of two opposite-sign leptons, back-to-back in the azimuthal angle φ and balanced in transverse momentum p T . In elastic interactions, the beam protons remain intact and escape undetected along the beam line. In the startup phase of the LHC, when the number of extra interactions per beam crossing (pileup) is small, these events can be produced exclusively, meaning that beyond the two signal leptons there is no additional detector activity above the noise threshold. A preliminary study 1 of the prospects for observing these processes with the CMS detector (described in detail elsewhere 2 ) has been performed; zero pileup is assumed throughout. While this study considers pp collisions at 14 TeV, similar γγ and γp simulation studies have been carried out in CMS for "ultraperipheral" Pb-Pb collisions at 5.5 TeV 3 .
The two-photon interaction γγ → ℓ + ℓ − is a nearly pure QED process, and can potentially be used for luminosity studies in the early LHC running. At high luminosity it will serve as a control sample for alignment of forward proton detectors 4 , and for studies of non-Standard Model physics 5,6,7,8,9,10,11 in γγ interactions. Beyond possible uses as a calibration and alignment sample, Υ photoproduction offers an opportunity to constrain models of QCD and diffraction. In particular, this process is sensitive to the generalised (or skewed) parton distribution functions (GPDs) of the proton. A measurement of Υ photoproduction at LHC would extend the coverage in the effective γp centre-of-mass energy by approximately one order of magnitude 12,13 .

Detector and event samples
The generated signal and background samples used in this study are processed through the full CMS analysis software unless otherwise noted. This includes detector simulation, trigger emulation, and offline reconstruction. The LP AIR 14 (two-photon) and ST ARLIGHT 15,16 (photoproduction) generators are used to generate the signal events. The largest source of irreducible background is dilepton production through inelastic (or dissociative) photon-exchange events. These events will appear exclusive when all products of the beam proton fragmentation escape detection. Although these events are expected to leave no activity within the central CMS hadronic calorimeter acceptance, they will often show activity in the CAST OR 17 or ZDC 18 forward calorimeters. These forward detector extensions can therefore be used to veto the inelastic events.
Other sources of dilepton backgrounds relevant to CMS include Drell-Yan production, quarkonium decays, heavy-flavor semileptonic decays, and W + W − production. The Pythia generator is used to generate all of these samples.

Event selection and background estimate
Starting from the sample of dileptons that satisfy the trigger requirements p T > 3 GeV/c (muons) and E T > 6 GeV/c (electrons), we require that the offline reconstruction find exactly two same flavor opposite-sign dileptons in the event. Further selections on the acoplanarity and transverse momentum balance of the leptons are applied to suppress backgrounds. In the µ + µ − channel, we require |∆φ(µ + µ − )| > 2.9 rad and |∆p T (µ + µ − )| < 2.0 GeV/c. In the e + e − channel, we require |∆φ(e + e − )| < 2.7 rad and |∆E T (e + e − )| < 5.0 GeV/c. A common exclusivity selection is applied for the dimuon and dielectron channels. The calorimeter exclusivity requirement is implemented by requiring that no more than 5 "extra" calorimeter towers have E > 5 GeV, where extra towers are defined as those separated from either of the lepton candidates by ∆R > 0.3 in the η − φ plane. A requirement that the 4 track multiplicity satisfy N (tracks) < 3 provides additional background suppression in the |η| < 2.5 region covered by the tracker. The distributions of all selection variables for the γγ → µ + µ − signal and for the sum of backgrounds are shown in Figure 2.  The remaining background is dominated by the inelastic photonexchange events. Generator-level acceptance studies show that approximately 2/3 of this background can be removed by applying a veto against activity in the CAST OR and ZDC detectors. The residual background from non-photon exchange processes is estimated by performing an exponential fit to the sideband of the extra calorimeter towers distribution, in the region 5 < N(towers) < 25. For an integrated luminosity of 100 pb −1 this procedure results in a background estimate of approximately 39 events. In the signal region, the non-photon exchange background is small compared to the inelastic photon-exchange contribution, even after the forward detector vetos are applied.
Due to the contribution of the theoretically less well understood inelastic events, the elastic signal cannot be extracted on an event-by-event basis. However, the ∆φ and ∆p T distributions provide a means of statistically separating the two contributions ( Figure 3). The Υ photoproduction signal can be extracted by performing a fit to the dimuon invariant mass distribution in the range 8 < m < 12 GeV/c 2 . The 1S, 2S, and 3S Υ resonances are fit to single Gaussians, while the sum of elastic and inelastic γγ → µ + µ − contributions are fit to a second order polynomial (Figure 4). For an assumed integrated luminosity of 100 pb −1 and the cross-section prediction from ST ARLIGHT , the three Υ resonances are clearly visible above the γγ continuum. A measurement of the t dependence of the cross-section is also possible; here t indicates the squared four-momentum transfer at the proton vertex. 6

Conclusions
With 100 pb −1 of integrated luminosity, a large sample of γγ → µ + µ − and γp → Υp → µ + µ − p events can be triggered and reconstructed in the CMS experiment, using a common selection for both samples. A smaller sample of γγ → e + e − events will also be collected. With minimal pileup these events can be distinguished using exclusivity requirements, and the inelastic backgrounds reduced using forward detector vetos. The Υ sample will allow measurements of cross-sections at significantly higher energies than previous experiments. The γγ → ℓ + ℓ − sample will serve as a calibration sample for studies of luminosity and lepton reconstruction.