Search for long-lived particles with displaced vertices in multijet events in proton-proton collisions at $\sqrt{s}=13\mathrm{TeV}$

Results are reported from a search for long-lived particles in proton-proton collisions at $\sqrt{s}=13\mathrm{TeV}$ delivered by the CERN LHC and collected by the CMS experiment. The data sample, which was recorded during 2015 and 2016, corresponds to an integrated luminosity of $38.5{\mathrm{fb}}^{-1}$. This search uses benchmark signal models in which long-lived particles are pair-produced and each decays into two or more quarks, leading to a signal with multiple jets and two displaced vertices composed of many tracks. No events with two well-separated high-track-multiplicity vertices are observed. Upper limits are placed on models of $R$-parity violating supersymmetry in which the long-lived particles are neutralinos or gluinos decaying solely into multijet final states or top squarks decaying solely into dijet final states. For neutralino, gluino, or top squark masses between 800 and 2600 GeV and mean proper decay lengths between 1 and 40 mm, the analysis excludes cross sections above 0.3 fb at 95% confidence level. Gluino and top squark masses are excluded below 2200 and 1400 GeV, respectively, for mean proper decay lengths between 0.6 and 80 mm. A method is provided for extending the results to other models with pair-produced long-lived particles.

Figure 1

Diagrams for the multijet (upper) and dijet (lower) benchmark signal models used in this analysis. In the multijet signal model, long-lived neutralinos (${\tilde{\chi }}^0$) or gluinos ($\tilde{g}$) decay into top, bottom, and strange antiquarks, via a virtual top squark ($\tilde{t}$). In the dijet signal model, long-lived top squarks decay into two down antiquarks. The charge conjugate processes are also considered.

Figure 2

Distribution of the distance between vertices in the $x\text{−}y$ plane, $d_{\mathrm{VV}}$, for simulated multijet signals with $m=800\mathrm{GeV}$, production cross section 1 fb, and $c\tau =0.3$, 1.0, and 10 mm, with the background template overlaid. All vertex and event selection criteria have been applied. The last bin includes the overflow events.

Figure 3

Signal efficiency as a function of signal mass and lifetime, for the multijet (upper) and dijet (lower) signal samples. All vertex and event selection criteria have been applied, as well as the requirement $d_{\mathrm{VV}}>0.4\mathrm{mm}$.

Figure 4

Distribution of $d_{\mathrm{BV}}$ in $\ge 5$-track one-vertex events for data and simulated multijet signals with $m=800\mathrm{GeV}$, production cross section 1 fb, and $c\tau =0.3$, 1.0, and 10 mm. Event preselection and vertex selection criteria have been applied. The last bin includes the overflow events.

Figure 5

Distribution of the distance between vertices in the $x\text{−}y$ plane in two-vertex events. The points show the data ($d_{\mathrm{VV}}$), and the solid lines show the background template ($d_{\mathrm{VV}}^{\mathrm{C}}$) normalized to the data, for events with two 3-track vertices (upper left), one 4-track vertex and one 3-track vertex (upper right), two 4-track vertices (lower left), and two $\ge 5$-track vertices (lower right). In each plot, the last bin includes the overflow events. The dotted lines indicate the boundaries between the three bins used in the fit.

Figure 6

Observed 95% C.L. upper limits on $\sigma {\mathcal{B}}^2$ for the multijet (left) and dijet (right) signals as a function of mass and mean proper decay length. The upper plots span $c\tau$ from 1 to 100 mm, and the lower plots span $c\tau$ from 0.1 to 1 mm. The overlaid mass exclusion curves assume gluino pair production cross sections for the multijet signals and top squark pair production cross sections for the dijet signals, and 100% branching fraction.

Figure 7

Observed and expected 95% C.L. upper limits on $\sigma {\mathcal{B}}^2$ for the multijet (left) and dijet (right) signals, as a function of mass for a fixed $c\tau$ of 0.3 mm (upper), 1.0 mm (middle), and 10 mm (lower). The gluino pair production cross section is overlaid for the multijet signals, and the top squark pair production cross section is overlaid for the dijet signals. The uncertainties in the theoretical cross sections include those due to the renormalization and factorization scales, and the parton distribution functions.

Figure 8

Observed and expected 95% C.L. upper limits on $\sigma {\mathcal{B}}^2$ for the multijet (left) and dijet (right) signals, as a function of $c\tau$ for a fixed mass of 800 GeV (upper), 1600 GeV (middle), and 2400 GeV (lower).