Physics Letters B 797 (2019) 134826 Contents lists available at ScienceDirect Physics Letters B www.elsevier.com/locate/physletb Evidence for light-by-light scattering and searches√ for axion-like particles in ultraperipheral PbPb collisions at sNN = 5.02 TeV .The CMS Collaboration CERN, Switzerland a r t i c l e i n f o a b s t r a c t Article history: Evidence for the light-by-light scattering process, γγ → γγ, in ultraperipheral PbPb collisions at a Received 10 October 2018 centre-of-mass energy per nucleon pair of 5.02 TeV is reported. The analysis is conducted using a Received in revised form 18 July 2019 −1 data sample corresponding to an integrated luminosity of 390 μb recorded by the CMS experiment Accepted 29 July 2019 at the LHC. Light-by-light scattering processes are selected in events with two photons exclusively Available online 1 August 2019 γ γ produced, each with transverse energy E > 2GeV, pseudorapidity |η | < 2.4, diphoton invariant mass Editor: M. Doser T γγ γγ m > 5GeV, diphoton transverse momentum pT < 1GeV, and diphoton acoplanarity below 0.01. After Keywords: all selection criteria are applied, 14 events are observed, compared to expectations of 9.0 ± 0.9 (theo) Light-by-light events for the signal and 4.0 ± 1.2(stat)for the background processes. The excess observed in data CMS relative to the background-only expectation corresponds to a significance of 3.7 standard deviations, UPC and has properties consistent with those expected for the light-by-light scattering signal. The measured Photoproduction fiducial light-by-light scattering cross section, σfid(γγ → γγ) = 120 ± 46 (stat) ± 28 (syst) ± 12 (theo) nb, PbPb γγ is consistent with the standard model prediction. The m distribution is used to set new exclusion limits on the production of pseudoscalar axion-like particles, via the γγ → a → γγ process, in the mass range ma = 5–90 GeV. © 2019 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license 3 (http://creativecommons.org/licenses/by/4.0/). Funded by SCOAP . 1. Introduction fluxes of quasireal photons emitted by the nuclei accelerated at TeV energies [8]. Ions accelerated at high energies generate strong Elastic light-by-light (LbL) scattering, γγ → γγ, is a pure electromagnetic fields, which, in the equivalent photon approxima- 2 2 quantum mechanical process that proceeds, at leading order in the tion [9–11], can be considered as γ beams of virtuality Q < 1/R , quantum electrodynamics (QED) coupling α, via virtual box dia- where R is the effective radius of the charge distribution. For lead ≈ grams containing charged particles (Fig. 1, left). In the standard (Pb) nuclei with radius R 7fm, the quasireal photon beams have 2 −3 2 model (SM), the box diagram involves contributions from charged virtualities Q < 10 GeV , but very large longitudinal energy (up ± fermions (leptons and quarks) and the W boson. Although LbL to Eγ = γ /R ≈ 80 GeV, where γ is the Lorentz relativistic factor), scattering via an electron loop has been indirectly tested through enabling the production of massive central systems with very soft the high-precision measurements of the anomalous magnetic mo- transverse momenta (pT 0.1GeV). Since each photon flux scales 2 ment of the electron [1] and muon [2], its direct observation in as the square of the ion charge Z , γγ scattering cross sections in 4 × 7 the laboratory remains elusive because of a very suppressed pro- PbPb collisions are enhanced by a factor of Z 5 10 compared 4 −9 duction cross section proportional to α ≈ 3 × 10 . Out of the to similar proton-proton or electron-positron interactions. two closely-related processes—photon scattering in the Coulomb Many final states have been measured in photon-photon in- field of a nucleus (Delbrück scattering) [3] and photon splitting teractions in ultraperipheral collisions of proton and/or lead + − in a strong magnetic field (“vacuum birefringence”) [4,5]—only the beams at the CERN LHC, including γγ → e e [12–21], γγ → + − former has been clearly observed [6]. However, as demonstrated W W [22–24], and first evidence of γγ → γγ reported by the in Ref. [7], the LbL process can be experimentally observed in ul- ATLAS experiment [25]with a signal significance of 4.4 standard traperipheral interactions of ions, with impact parameters larger deviations (3.8 standard deviations expected). The final-state signa- than twice the radius of the nuclei, exploiting the very large ture of interest in this analysis is the exclusive production of two (∗) (∗) photons, PbPb → γγ → Pb γγPb , where the diphoton final state is measured in the otherwise empty central part of the detec- E-mail address: cms -publication -committee -chair @cern .ch. tor, and the outgoing Pb ions (with a potential electromagnetic ex- https://doi.org/10.1016/j.physletb.2019.134826 0370-2693/© 2019 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Funded by SCOAP3. 2 The CMS Collaboration / Physics Letters B 797 (2019) 134826 + − Fig. 1. Schematic diagrams of light-by-light scattering (γγ → γγ, left), QED dielectron (γγ → e e , centre), and central exclusive diphoton (gg → γγ, right) production in ∗ ultraperipheral PbPb collisions. The ( ) superscript indicates a potential electromagnetic excitation of the outgoing ions. (∗) citation denoted by the superscript) survive the interaction and (L1), composed of custom hardware processors, uses information escape undetected at very low θ angles with respect to the beam from the calorimeters and muon detectors to select the most in- direction (Fig. 1, left). The dominant backgrounds are the QED teresting events in a fixed time interval of less than 4μs. The + − production of an exclusive electron-positron pair (γγ → e e ) high-level trigger (HLT) processor farm further decreases the event ± where the e are misidentified as photons (Fig. 1, centre), and rate before data storage. A more detailed description of the CMS gluon-induced central exclusive production (CEP) [26]of a pair of detector, together with a definition of the coordinate system used photons (Fig. 1, right). and the relevant kinematic variables, can be found in Ref. [36]. The γγ → γγ process at the LHC has been proposed as a par- ticularly sensitive channel to study physics beyond the SM. Modifi- 3. Simulation and reconstruction cations of the LbL scattering rates can occur if, e.g. new heavy par- ticles, such as magnetic monopoles [27], vector-like fermions [28], The light-by-light signal is generated with the MadGraph or dark sector particles [29], contribute to the virtual corrections v5 [37]Monte Carlo (MC) event generator, with the modifications of the box depicted in Fig. 1. Other new spin-even particles, such discussed in Refs. [7,38]to include the nuclear photon fluxes and as axion-like particles (ALPs) [30]or gravitons [31], can also con- the elementary LbL scattering process. The latter includes all quark tribute to the LbL scattering continuum or to new diphoton res- and lepton loops at leading order, but omits the W boson contribu- γγ onances. In addition, light-by-light cross sections are sensitive to tions, which are only important for diphoton masses m > 2mW. Born–Infeld extensions of QED [32], and anomalous quartic gauge Next-to-leading order (NLO) quantum chromodynamics and QED couplings [33]. corrections increase σγγ→γγ by just a few percent [39] and are + − We report a study of the γγ → γγ process, using PbPb also neglected here. Exclusive γγ → e e events can be misiden- √collision data recorded by the CMS experiment in 2015 at tified as LbL scattering if neither electron track is reconstructed or = sNN 5.02 TeV and corresponding to an integrated luminosity if both electrons undergo hard bremsstrahlung. This QED process −1 of 390 μb . A comparison of exclusive diphoton and dielectron is generated using the starlight v2.76 [40,41]event generator, yields, with almost identical event selection and reconstruction also based on the equivalent-photon fluxes. Since the cross section + − efficiencies, is provided as a function of key kinematic variables for the QED e e background is four to five orders of magnitude to check of the robustness of the analysis. The ratio of the LbL larger than that for LbL scattering, and it relies on physics objects + − scattering to QED e e production cross sections is reported, (electrons) that are very similar to those of the signal (photons), so as to reduce the dependence on various experimental correc- the exclusive dielectron background is analysed in depth in order γγ tions and uncertainties. Using the measured m distribution, new to estimate many of the (di)photon efficiencies directly from the + − exclusion limits are set on ALP production, in the mass range data, as well as to determine an LbL/(QED e e ) production cross ma = 5–90 GeV. sections ratio with reduced common uncertainties. The central ex- clusive production process, gg → γγ, is simulated using superchic 2. The CMS detector 2.0 [42], where the computed proton-proton cross section [26]is 2 4 conservatively scaled to the PbPb case by multiplying it by A Rg , The central feature of the CMS apparatus is a superconduct- where A = 208 is the mass number of lead and Rg ≈ 0.7is a gluon ing solenoid of 6 m internal diameter, providing a magnetic field shadowing correction in the relevant kinematic range [43], and of 3.8T. Within the solenoid volume are a silicon pixel and strip where the rapidity gap survival probability, encoding the probabil- tracker, a lead tungstate crystal electromagnetic calorimeter (ECAL), ity to produce the diphoton system exclusively without any other and a brass and scintillator hadron calorimeter (HCAL), each com- hadronic activity, is assumed to be 100%.