Search for long-lived charginos and $τ$-sleptons using final states with a disappearing track in $pp$ collisions at $\sqrt{s} = 13$ TeV with the ATLAS detector

Abstract

This paper reports a search for decays of long-lived charginos or $τ$-sleptons to final states containing a short disappearing track, a single high-energy jet, and missing transverse momentum. The search uses 137 fb$^{-1}$ of data from 13 TeV proton-proton collisions recorded by the ATLAS detector during Run 2 of the LHC. Multiple search regions are defined, all requiring the presence of a track reconstructed from either three or four measurements in the innermost layers of the ATLAS detector. Regions with tracks having only three measurements are further characterised by the absence or presence of a low-energy charged pion reconstructed using a dedicated algorithm, leveraging machine learning. Data-driven methods are used to estimate the background contributions in the search regions. No significant excesses are found and 95% CL lower limits are placed on the masses of charginos and $τ$-sleptons in the lifetime range $0.01{-}10$ ns. Observed (expected) mass limits of up to 225 GeV (250 GeV) are set for pure-higgsino charginos in scenarios with lifetimes below 0.03 ns, where the electroweakino mass splitting is entirely due to loop corrections involving the Standard Model bosons, and up to 720 GeV (840 GeV) for charginos with a lifetime of around 1 ns. For wino production, charginos with masses up to 880 GeV (1020 GeV) are excluded for lifetimes of around 1 ns. For $τ$-sleptons with lifetimes of around 1 ns, masses are excluded up to 320 GeV (390 GeV) in Constrained Minimal Supersymmetric Standard Model scenarios and 300 GeV (380 GeV) in Gauge-Mediated Supersymmetry-Breaking scenarios.

Figures (14)

• Figure 1: Representative signal diagrams for \ref{['fig:intro:feynman:c1n1']} the electroweak production of $\hbox{$\tilde{\chi}^\pm_1$}\xspace\hbox{$\tilde{\chi}^0_1$}\xspace$, and $\tau$-slepton pair production in \ref{['fig:intro:feynman:ststCMSSM']} CMSSM-inspired or \ref{['fig:intro:feynman:ststGMSB']} GMSB-inspired SUSY models. The signal signature consists of a long-lived $\hbox{$\tilde{\chi}^\pm_1$}$ or $\tau$-slepton, missing transverse momentum, and quarks or gluons, which are observed as jets ($j$). When considering the electroweakino scenarios, the production of $\hbox{$\tilde{\chi}^+_1$}\xspace\hbox{$\tilde{\chi}^-_1$}\xspace$ is also considered, as is $\hbox{$\tilde{\chi}^\pm_1$}\xspace\hbox{$\tilde{\chi}^0_2$}\xspace$ specifically in the higgsino scenario.
• Figure 2: Representative distributions related to the reconstruction of tracklets and charged pions: \ref{['fig:reco:TrackEff']} event yields and reconstruction efficiencies for three- and four-layer tracklet reconstruction as a function of $\hbox{$\tilde{\chi}^\pm_1$}$ decay radius; \ref{['fig:reco:pionBDT']} BDT output score used to judge if an event contains a low-energy pion, showing the training dataset (solid histogram) and the test dataset (points); \ref{['fig:reco:pionEff']} event yields and reconstruction efficiency for charged pions as a function of the true pion transverse momentum as measured in the lab-frame, where the acceptance includes all charged-pion reconstruction effects but not geometric acceptance from tracklet reconstruction (which is assumed to already have been reconstructed). The true transverse momentum is defined as $p_{\text{T}}\xspace = \sqrt{p_x^2 + p_y^2}$ in the global detector coordinate system with the $z$-axis along the beam pipe. For \ref{['fig:reco:TrackEff']} and \ref{['fig:reco:pionEff']} the filled histogram displays the total yield without applying a selection, while the dashed lines present the yield after applying the tracklet selection or tracklet+pion selection. The points denote the reconstruction efficiency. A representative higgsino signal scenario with $m(\hbox{$\tilde{\chi}^\pm_1$}\xspace) = 250$ GeV and $\tau(\hbox{$\tilde{\chi}^\pm_1$}\xspace)= 0.04$ ns is shown.
• Figure 3: Schematic diagram illustrating how the signal scenarios are detected and how the background processes may enter the SR selections. The detector layers are not shown to scale. Taken from Ref. SUSY-2018-19.
• Figure 4: Individual components used in the estimation of the fake background: \ref{['fig:bkg:34_layer_Template']} shows the original template, $N^{\mathrm{TRfake}}(p_{\text{T}}\xspace)$, as a solid line. The templates after the application of the transfer factors are shown as dashed lines; \ref{['fig:bkg:fake_SF_d0']} shows the $|d_{0}|/\sigma(d_{0})$ dependence introduced when extrapolating over $|z_0\sin\theta|$ and the binwise, with respect to $|d_{0}|/\sigma(d_{0})$, transfer factors $\mathrm{TF}^{\mathrm{fake}}_{d_0\textrm{-correction}}(d_0)$ used to remove the dependence; and \ref{['fig:bkg:34_Hybrid']} shows a comparison between the pure fake and hybrid fake contributions and the $p_{\text{T}}$-dependent transfer factors $\mathrm{TF}^{\mathrm{fake}}_{\mathrm{hybrid}}(p_{\text{T}}\xspace)$ used to extrapolate from the template region that contains pure fakes to the SR, which contains a mixture of hybrid and pure fake components.
• Figure 5: Individual contributions used in the estimation of the electron background: \ref{['fig:bkg:el_Template']} presents the original electron $p_{\text{T}}$ template, $N^{\mathrm{TRe}}(p_{\text{T}}\xspace)$, as a solid line, with the templates also shown after the application of each transfer/smearing factor as dashed lines; \ref{['fig:bkg:el_TF_trk']} shows the transfer factor, $\mathrm{TF}^e_{\mathrm{tracklet}}$, used to correct for the likelihood of reconstructing an electron track as a tracklet; \ref{['fig:bkg:el_TF_calo']} shows the transfer factors, $\mathrm{TF}^{e}_{\textrm{calo-veto}}$, used to correct for the likelihood of an electron not being associated with a calorimeter energy cluster; and \ref{['fig:bkg:el_smear']} shows the smearing functions, $f^{e}_{\mathrm{smear}}$, used to correct the measured electron track $p_{\text{T}}$ to the tracklet $p_{\text{T}}$ (defined on the figure). While the tracklet $p_{\text{T}}$ template and the TFs are dependent upon both $p_{\text{T}}$ and $\eta$, the components are shown with their dependence upon the tracklet $p_{\text{T}}$ distribution as a representative distribution.