Dimuons in neutrino telescopes: New predictions and first search in IceCube

Neutrino telescopes allow powerful probes of high-energy astrophysics and particle physics. Their power is increased when they can isolate different event classes, e.g., by flavor, though that is not the only possibility. Here we focus on a new event class for neutrino telescopes: dimuons, two energetic muons from one neutrino interaction. We make new theoretical and observational contributions. For the theoretical part, we calculate dimuon-production cross sections and detection prospects via deep-inelastic scattering (DIS; where we greatly improve upon prior work) and $W$-boson production (WBP; where we present first results). We show that IceCube should have $\simeq 130$ dimuons ($\simeq 6$ from WBP) in its current data and that IceCube-Gen2, with a higher threshold but a larger exposure, could detect $\simeq 620$ dimuons ($\simeq 30$ from WBP) in 10 years. These dimuons are almost all produced by atmospheric neutrinos. For the observational part, we perform a simple but conservative analysis of IceCube public data, finding 19 candidate dimuon events. Subsequent to our paper appearing, visual inspection of these events by the IceCube Collaboration revealed that they are not real dimuons, but instead arise from an internal reconstruction error that identifies some single muons crossing the dust layer as two separate muons. To help IceCube and the broader community with future dimuon searches, we include the updated full details of our analysis. Together, these theoretical and observational contributions help open a valuable new direction for neutrino telescopes, one especially important for probing high-energy QCD and new physics.

Figure 1

Examples of dimuons and other events with similar signatures. Dimuon signals may be either starting and throughgoing. Backgrounds from two coincident single muons are rare except for downgoing events. New-physics signals, e.g., double staus produced in supersymmetric models [47, 48, 49, 50, 51, 52, 53, 54], can have a similar topology.

Figure 2

The dominant channel for dimuon production: a strange-quark-induced DIS event between ${\nu }_{\mu }$ and a nucleon. The $D$ meson is from hadronization of the charm quark, and $X$ indicates hadronic particles that produce a shower. A similar diagram holds for ${\overline{\nu }}_{\mu }$.

Figure 3

A dimuon event traveling in a high-energy neutrino detector. The vertical lines and the dots on them represent the strings in the detector and the DOMs on them, respectively. In our calculation, we require $\mu 1$ and $\mu 2$ to separate by at least $2D_{\mathrm{v}}$ to be distinguishable, where $D_{\mathrm{v}}$ is the vertical spacing between DOMs (here we show a larger separation).

Figure 4

Left: Cross sections on ${\mathrm{H}}_2\mathrm{O}$ targets for DIS (red, $[{\nu }_{\mu }+{\overline{\nu }}_{\mu }]/2$) from our calculation here and WBP (blue, ${\nu }_{\mu }$ or ${\overline{\nu }}_{\mu }$) from our previous work [20, 21]. Solid: Total cross sections. Dashed: Dimuon cross sections. Dotted: Dimuon cross sections with cuts of $E_{\mu 2}>E_{\mathrm{th}}$ and $R_{\mu 2}{\theta }_{\mu \mu }>2D_{\mathrm{v}}$ for starting muons (Sec. 2b2) in IceCube. Dot-dashed: same, but for IceCube-Gen2. Right: Parent neutrino spectra for starting muons (10 years of exposure, same line styles as the cross sections). For throughgoing muons, the cut cross sections and parent neutrino spectra (not shown) are more involved but are qualitatively similar.

Figure 5

Our predicted dimuon spectra in IceCube. Left: Starting dimuons ($\simeq 37$ from DIS and $\simeq 0.3$ from WBP). Right: Throughgoing dimuons ($\simeq 85$ from DIS and $\simeq 6.0$ from WBP). We define $\mu 1$ to be the more energetic of the two muons.

Figure 6

Same as Fig. 5, but for 10 years of the IceCube-Gen2 optical array. Left: Starting events ($\simeq 370$ from DIS and $\simeq 5.8$ from WBP). Right: Throughgoing events ($\simeq 230$ from DIS and $\simeq 22$ from WBP).

Figure 7

Energy distributions of the 19 dimuon candidates from the IceCube data. Left: Scatter plot of the energies for each muon in the dimuon event. The gray regions are disfavored by either energy threshold (we estimate 400 GeV from the energy distribution of muon events in the IceCube dataset) or the definition $E_{\mu 1}\ge E_{\mu 2}$. Right: Binned distributions of $E_{\mu 1}$ (upper) and $E_{\mu 2}$ (lower), compared to our prediction (solid line) as well as the distribution of single-muon events, as if the dimuon candidates were somehow misreconstructed single muons. See text for details.

Figure 8

Angular-separation distribution of the 19 dimuon candidates and our prediction, which is dominated by throughgoing DIS events (see Table 1). The dropoff at low ${\theta }_{\mu \mu }$ is primarily due to the angular cut (see Sec. 2b2), while the dropoff at large ${\theta }_{\mu \mu }$ is caused by the kinematics, which are affected by the energy threshold. We show results without and with angular-resolution smearing (see text); we show a representative scale for the effective uncertainty.

Figure 9

Distributions of parton-momentum fraction $x$ and factorization scale $Q$ for dimuon events. Those in blue are for our predicted dimuons (numbers of events in $x-Q$ bins) as titled in each panel, corresponding to the dimuons in Figs. 5 and 6. All panels use the same color scale. The dots are the data points of CCFR dimuons used for CT18 PDF set [61, 64, 68].

Figure 10

Starting dimuons for 10 years of IceCube-Gen2 with their shower energies below different thresholds, as noted in each panel. $\mathrm{\Delta }{\mathrm{E}}_{\mu \mu }(300\mathrm{m})$ is the average energy lost by the two muons in their initial path length of 300 m. The top left panel has a threshold of infinity and is the same as the left panel of Fig. 6. The total number of starting WBP dimuons (signal) is about 5.8, and they are all showerless. The total numbers of starting DIS dimuons (background) are about 370, 4.4, 0.4, and 0.1 for the top left, top right, bottom left, and bottom right panels, respectively.

Figure 11

Distributions in RA and $\cos {\theta }_z$ of the 19 dimuon candidates, along with our predictions, where $\mathrm{RA}=(\mathrm{RA}1+\mathrm{RA}2)/2$ and $\cos {\theta }_z=\cos ([{\theta }_{z,1}+{\theta }_{z,2}]/2)$, where ${\theta }_{z,1/2}\simeq \text{Dec}1/2+90°$ for IceCube’s location. The data agree well with predictions.