PaperlayerView exact revision on arXiv

Source-linked AI summary

Search for resonant and nonresonant new phenomena in high-mass dilepton final states at $\sqrt{s} = $ 13 TeV

CMS Collaboration

arXiv:2103.02708v2hep-ex

TL;DR

The paper searches high-mass dielectron and dimuon spectra for resonant and nonresonant physics beyond the SM using up to 140 fb−1 of 13 TeV CMS data. It combines resonance fits, binned nonresonant tests, and a lepton-flavor-universality comparison, finding no significant deviation from SM expectations while setting limits on several new-physics models.

  • Problem

    High-mass dilepton spectra may reveal resonant or nonresonant phenomena beyond the SM, including possible lepton flavor universality violation.

  • Method

    CMS analyzes 2016–2018 dielectron and dimuon data with resonant mass-window fits, binned nonresonant searches, and a dimuon-to-dielectron cross-section-ratio test.

  • Results

    No significant deviation from SM background expectations is observed, while limits are set on narrow resonances, contact interactions, and ADD large-extra-dimension models.

  • Takeaways & Limitations

    The results provide stringent constraints on spin-1 and spin-2 resonances, dark-matter mediators, contact interactions, and large extra dimensions, with no significant lepton flavor universality deviation.

  • Takeaways & Limitations

    For the largest considered resonance width, 10%, background-parameterization choices can significantly bias low-mass limits, so results below 700 GeV are not shown.

Abstract

from arXiv · show

A search is presented for physics beyond the standard model (SM) using electron or muon pairs with high invariant mass. A data set of proton-proton collisions collected by the CMS experiment at the LHC at $\sqrt{s} =$ 13 TeV from 2016 to 2018 corresponding to a total integrated luminosity of up to 140 fb$^{-1}$ is analyzed. No significant deviation is observed with respect to the SM background expectations. Upper limits are presented on the ratio of the product of the production cross section and the branching fraction to dileptons of a new narrow resonance to that of the Z boson. These provide the most stringent lower limits to date on the masses for various spin-1 particles, spin-2 gravitons in the Randall--Sundrum model, as well as spin-1 mediators between the SM and dark matter particles. Lower limits on the ultraviolet cutoff parameter are set both for four-fermion contact interactions and for the Arkani-Hamed, Dimopoulos, and Dvali model with large extra dimensions. Lepton flavor universality is tested at the TeV scale for the first time by comparing the dimuon and dielectron mass spectra. No significant deviation from the SM expectation of unity is observed.

1 Introduction

The paper searches high-mass dilepton spectra for resonant and nonresonant phenomena beyond the SM using the full 2016–2018 CMS dataset, while also testing lepton flavor universality.

  • Motivation: The analysis targets spin-1 and spin-2 resonances, nonresonant high-mass lepton-pair production, and possible lepton flavor universality violation.The motivation includes new gauge bosons, extra-dimensional gravitons, dark-matter mediators, contact interactions, and flavor anomalies.
  • Dataset and scope: The paper significantly extends previous CMS results with the full 13 TeV dataset from 2016–2018, corresponding to 137 fb−1 in dielectron and 140 fb−1 in dimuon events.The dimuon sample includes additional data recorded under conditions unsuitable for electron reconstruction.
  • Search strategy: Resonant searches scan mass and width hypotheses with unbinned likelihood fits in mass windows, using data to determine background normalization.The window size depends on the assumed intrinsic width and mass-dependent detector resolution; limits use the resonance-to-Z cross-section and branching-fraction ratio.
  • Search strategy: Nonresonant searches combine counting experiments across dilepton-mass and Collins–Soper scattering-angle bins, retaining lower-mass information about signal–background interference.The highest-mass bins provide most sensitivity because their signal-to-background ratio is more favorable.
  • Lepton flavor universality: Lepton flavor universality is tested for the first time in these final states by comparing the dimuon-to-dielectron differential cross-section ratio with the SM expectation of unity.The comparison includes corrections for detector effects, acceptances, and lepton efficiencies.

2 The CMS detector

The CMS detector combines tracking, calorimetry, and muon systems inside a 3.8 T solenoid, with a two-tier trigger reducing collision rates for storage.

  • Detector components: A 3.8 T superconducting solenoid encloses the inner tracker, electromagnetic calorimeter, and hadron calorimeter, while the muon system lies outside the solenoid.The tracker covers |η| < 2.5, calorimeters extend through the forward region, and muon detection covers |η| < 2.4.
  • Trigger system: The two-tier trigger uses custom hardware at about 100 kHz and a high-level software trigger to reduce the event rate to about 1 kHz before storage.The first-level decision is made within less than 4 µs.

3 Signal models

The paper models both resonant and nonresonant high-mass dilepton signatures from several extensions of the Standard Model, including new gauge bosons, extra-dimensional gravitons, dark-matter mediators, contact interactions, and virtual gravitons.

  • 3.1 Models with resonant signatures: Three resonant benchmarks are considered: spin-1 Z′ bosons from U′(1) models, spin-2 Randall–Sundrum gravitons, and spin-1 dark-matter mediators.The U′(1) benchmark classes include GSM, LRS, and E6 models; their production depends on model-specific quark couplings and parton distributions.
  • 3.1 Models with resonant signatures: The benchmark Z′ models are characterized by mixing-dependent couplings, branching fractions, and width-to-mass ratios summarized across GSM, LRS, and E6 classes.In the narrow-width approximation, the production cross section is expressed through up- and down-type quark couplings and their parton distribution functions.
  • 3.1 Models with resonant signatures: The dark-matter benchmark assumes a single dark-matter particle coupled through either a vector or axial-vector spin-1 mediator.The considered coupling choices include a vector mediator with small lepton couplings and an axial-vector mediator with equal quark and lepton couplings; mediator interference is treated as negligible.
  • 3.2 Models with nonresonant signatures: Two nonresonant models are studied: virtual spin-2 gravitons in the ADD model and four-fermion contact interactions associated with fermion substructure.The contact-interaction search treats LL, RR, and LR/RL helicity currents separately, with potentially different compositeness scales.
  • 3.2 Models with nonresonant signatures: The dilepton cross section is written as the Standard Model Drell–Yan contribution plus interference and pure-signal terms, with interference potentially constructive or destructive.For ADD production, the ultraviolet cutoff conventions use ΛT, MS, or MS together with the number of extra dimensions; the effective theory is reliable only below its applicable energy scale.

4 Simulated event samples

The analysis uses year-specific simulated samples for backgrounds and signals, with generator, parton-distribution, higher-order, detector, pileup, and trigger corrections adapted to changing data-taking conditions.

  • 4 Simulated event samples: Dedicated simulated samples are produced for each of the three data-taking years to reflect changing beam and detector conditions.Different event generators are used for individual background and signal processes.
  • 4 Simulated event samples: The dominant Drell–Yan background is generated with POWHEG at NLO, showered with PYTHIA, and corrected for NNLO QCD and missing NLO electroweak effects.The corrections use mass-dependent factors obtained with FEWZ and photon-inclusive PDFs.
  • 4 Simulated event samples: Top-quark and diboson backgrounds are simulated with POWHEG, PYTHIA, or MADGRAPH5_aMC@NLO and normalized using higher-order cross sections where specified.The top-pair sample uses an NNLO normalization, while single-top tW uses a cross section computed to next-to-next-to-leading-logarithmic accuracy.
  • 4 Simulated event samples: High-mass signal samples include a 5 TeV Z′SSM sample, Randall–Sundrum graviton samples spanning 250–4000 GeV, and LO CI and ADD samples with year-dependent generation settings.Signal and Drell–Yan samples are simulated simultaneously for the nonresonant models to account for interference, while mass reweighting aligns 2017–2018 distributions with 2016 samples.
  • 4 Simulated event samples: Detector-related corrections account for pileup and the 2016–2017 ECAL trigger-prefiring inefficiency in the dielectron channel.The detector response is simulated with GEANT4, and simulated dielectron events are reweighted for both pileup and trigger inefficiency.

5 Lepton reconstruction and event selection

CMS reconstructs and selects high-mass dielectron and dimuon events with channel-specific trigger, charge, and detector-acceptance requirements, then models their mass-dependent acceptance and efficiency.

  • 5.1 Electron reconstruction and selection: Electrons require pT > 35 GeV and |ηC| < 1.44 or 1.57 < |ηC| < 2.50, excluding the lower-quality barrel–endcap transition region.At least one electron must be in the barrel to suppress W+jets and QCD multijet backgrounds from endcap misreconstruction.
  • 5.1 Electron reconstruction and selection: Electron charges are not required to be opposite because high-energy charge misidentification rises from about 1% near the Z peak to approximately 10% at several TeV.Same-sign events are retained and their charge ambiguity is resolved using the lower-ηC electron assumption described in the selection.
  • 5.1 Electron reconstruction and selection: At 1 TeV, electron-pair efficiency is 72% (2016) or 68% (2017–2018) for barrel-barrel events, and 67%, 60%, or 67% for barrel-endcap events.The efficiency remains stable within a few percent at larger masses; lower values reflect trigger changes and trigger prefiring.
  • 5.2 Muon reconstruction and selection: Dimuon events use single-muon triggers, require mμμ > 150 GeV and opposite charges, and include a backup trigger that raises efficiency above 1 TeV by approximately 1%.Muon reconstruction incorporates dedicated algorithms for pT > 200 GeV and includes data recorded during degraded calorimeter conditions.
  • 5.2 Muon reconstruction and selection: The acceptance–efficiency product is simulated separately for dielectron and dimuon pairs and for spin-1 and spin-2 signals, then parameterized as a function of mass.Above 200 GeV, reconstruction and selection efficiencies are nearly mass-independent; acceptance changes reflect lepton angular distributions, with a slight high-mass RS-graviton drop from changing production modes.

6 Background estimation

The background model is dominated by Drell–Yan production, with other prompt processes taken mainly from simulation and jet-misidentification backgrounds estimated from data control regions.

  • 6 Background estimation: Drell–Yan production is the dominant irreducible background in the high-mass dilepton selection.Additional backgrounds include tt, tW, diboson, QCD multijet, and W+jets events reconstructed as dileptons.
  • 6 Background estimation: Drell–Yan and most other backgrounds are estimated from simulation, while jets or nonisolated leptons misidentified as leptons are evaluated using data control regions.The Drell–Yan shape uses POWHEG samples with mass-dependent higher-order corrections; the DY → ττ contribution uses MADGRAPH5_aMC@NLO without such corrections because it is small.
  • 6 Background estimation: The combined background shape is normalized to data in the Z-boson control region 60 < mℓℓ < 120 GeV.For 2017 and 2018, the DY simulation is reweighted to improve the low-pT Z-boson description before normalization factors are calculated.

7 Systematic uncertainties

Systematic uncertainties are evaluated separately for resonant and nonresonant searches, including detector, background, theoretical, and simulation-related effects.

  • Dimuon mass-resolution uncertainties are 15% in 2016 and 8.5% in 2017–2018, while dielectron resolution uncertainty is negligible.
  • High-mass theoretical uncertainties in the Drell–Yan cross section grow to 20% at mℓℓ = 6 TeV, mainly from higher-order corrections and PDFs.
  • Jet-misidentification backgrounds carry a 50% uncertainty in both channels, although further extrapolation effects have negligible impact on results.
  • For nonresonant signals, PDF reweighting uncertainty reaches 30% at 5 TeV and changes limits on Λ and ΛT by up to 5%.
  • Resonant limits use Z-peak and resonance-window normalization, reducing common uncertainties and retaining mass-dependent signal-modeling uncertainties.

8 Results

The combined 2016–2018 dilepton spectra and background yields show agreement between data and SM expectations, with distributions also examined by scattering angle for nonresonant searches.

  • Data agree with expected backgrounds overall, despite a slight dielectron excess above 1.8 TeV; four events have masses above 3 TeV.The four measured masses are 3.35 and 3.47 TeV in dielectrons and 3.07 and 3.34 TeV in dimuons.
  • Figure 2 shows electron and muon invariant-mass spectra, data-to-background ratios, uncertainty bands, and illustrative GKK and Z′ SSM signals.
  • Figure 3 separates electron and muon spectra into cos θ∗ < 0 and cos θ∗ ≥ 0 bins to support the nonresonant search.
  • Table 4 compares observed and expected dielectron and dimuon yields across mass ranges, including total backgrounds and their three main categories.

9 Statistical interpretation

The analysis sets confidence-level limits for resonant and nonresonant dilepton signals using Bayesian statistical procedures and channel combinations. No significant departure from the Standard Model is established, while strong constraints are obtained for resonance, extra-dimension, contact-interaction, and flavor-universality models.

  • 9.1 Search for resonant signals: Spin-2 resonance mass limits range from slightly over 2 to almost 5 TeV for k/M_Pl values of 0.01, 0.05, and 0.10, improving previous CMS limits by about 370–530 GeV.The stronger low-mass sensitivity results from more central lepton production and increased detector acceptance.
  • 9.1 Search for resonant signals: Mediator masses below 1.92 TeV in the vector model and 4.64 TeV in the axial-vector model are excluded for large dark-matter masses.The limits are strongest when dark-matter decays are suppressed, narrowing the mediator width and enhancing decays to leptons.
  • 9.2 Search for nonresonant signals: ADD ultraviolet-cutoff limits reach 7.5 TeV in the combined GRW convention, while the full considered range is 5.9–8.9 TeV and improves previous limits by 0.5–1.0 TeV.The n = 2 HLZ convention is omitted because gravitational and astrophysical bounds already exceed this analysis’s reach.
  • 9.2 Search for nonresonant signals: Contact-interaction cutoff limits range from 23.9 to 36.4 TeV, with angular binning improving limits by approximately 1 TeV for destructive and 3 TeV for constructive interference.The electron-channel observed limit is weakened by up to two standard deviations because of the high-mass electron excess.
  • 9.3 Search for lepton flavor universality violation: The dimuon-to-dielectron differential cross-section ratio agrees with unity up to 1.5 TeV, while larger-mass deviations are associated with the slight dielectron excess and large statistical uncertainties.The unfolded particle-level ratio provides a basis for testing models that predict lepton flavor universality violation.

10 Summary

The CMS search finds no significant deviation from SM expectations while setting stringent constraints on resonant and nonresonant new-physics models and testing lepton flavor universality at the TeV scale.

  • The search sets upper limits for new resonances and finds the most stringent-to-date lower mass limits for spin-1 and spin-2 resonances.The spin-2 Randall–Sundrum graviton limits span 2.47–4.78 TeV for k/MPl values from 0.01 to 0.1.
  • Mediator masses below 1.92 (4.64) TeV are excluded for vector (axial-vector) dark-matter mediators at large dark-matter mass.For mDM = 0, the corresponding limits are 1.04 and 3.41 TeV.
  • 23.8–36.4 TeV lower limits are obtained on the ultraviolet cutoff Λ for four-fermion contact interactions, depending on helicity and interference sign.
  • 5.9–8.9 TeV lower limits are obtained on the ultraviolet cutoff in the ADD large-extra-dimensions model, depending on parameter convention.
  • The dimuon and dielectron spectra are compared at the TeV scale after detector corrections, with no significant deviation from lepton flavor universality observed.Figure 12 presents the ratio Rµ+µ−/e+e− versus dilepton mass for barrel, endcap, and combined events.
Loading 2103.02708v2…