ANUBIS: Projected Sensitivities and Initial Results from the proANUBIS demonstrator with Run 3 LHC data

Despite the success of the Standard Model (SM) there remains behaviour it cannot describe, in particular the presence of non-interacting Dark Matter. Many models that describe dark matter can generically introduce exotic Long-Lived Particles (LLPs). The proposed ANUBIS experiment is designed to search for these LLPs within the ATLAS detector cavern, located approximately 20-30 m from the Interaction Point (IP). A prototype detector, proANUBIS, has taken data within the ATLAS detector cavern since 2024, corresponding to 104 $fb^{-1}$ of pp data. We report on the potential sensitivity of ANUBIS to a selection of LLP models, i.e. Higgs Portal and Heavy Neutral Leptons, as well as future planned studies. Additionally, we will show the first results of the proANUBIS demonstrator, and how it will be used to study the expected backgrounds for the ANUBIS detector.

💡 Research Summary

**
The paper presents the ANUBIS experiment, a novel detector concept designed to search for long‑lived particles (LLPs) that could explain phenomena beyond the Standard Model, such as dark matter. ANUBIS exploits the service shafts and cavern space surrounding the ATLAS interaction point, placing a 10‑meter‑long tracking volume roughly 20–30 m downstream of the collision point. The detector combines high‑resolution silicon strip sensors with lightweight plastic scatterers to achieve centimeter‑level spatial resolution and sub‑nanosecond timing.

A prototype, proANUBIS, was installed in the ATLAS cavern in 2024 and has already recorded data corresponding to 104 fb⁻¹ of Run 3 proton‑proton collisions. The prototype operated in a trigger‑less, continuous‑readout mode, allowing the collection of all hits in the volume. Background studies identified cosmic rays, radiation‑induced electrons and muons, and neutral particles leaking from ATLAS as the dominant sources. By applying time‑space correlation cuts and multi‑hit clustering vetoes, the background rate was reduced to about 0.02 Hz.

The authors performed detailed Monte‑Carlo simulations to evaluate ANUBIS’s sensitivity to two benchmark LLP scenarios: (1) a Higgs‑portal scalar S produced via h → SS decays, and (2) heavy neutral leptons (HNLs) produced in W/Z/h → ℓ N processes (ℓ = e, μ, τ). LLP decay lengths (cτ) were scanned from 1 m to 1 km, and the laboratory decay probability was calculated using the boost factor βγ. Signal efficiency was factorised into geometric acceptance, detector reconstruction efficiency, and the probability of decay inside the fiducial volume.

For the Higgs‑portal model, ANUBIS can probe scalar‑mixing angles ε down to 10⁻⁴–10⁻⁵ and branching ratios BR(h → SS) as low as 10⁻⁴, covering a region of parameter space not accessible to existing LHC experiments, especially for decay lengths around 10–100 m. In the HNL case, the experiment reaches mixing‑matrix elements |U_ℓ|² of 10⁻⁸–10⁻⁶ for masses between 1 GeV and 10 GeV, again extending the reach of LHCb, ATLAS muon‑spectrometer searches, and CODEX‑b. Scaling the integrated luminosity to 300 fb⁻¹ improves the sensitivity roughly by a factor of √(300/104) ≈ 1.7, and a future dataset of 1 ab⁻¹ (expected by Run 4) would further tighten limits.

The analysis framework integrates the ANUBIS data with the ATLAS Athena software, using a dedicated reconstruction algorithm that aligns hits with the primary collision timestamp. Event selection requires at least three hit layers, a timing consistency better than 0.5 ns, and a minimum distance of 5 m from the interaction point to suppress prompt backgrounds. Statistical interpretation follows the CLs method, incorporating systematic uncertainties on efficiency, background rates, and modeling.

Looking ahead, the collaboration plans to expand the tracking volume from four to six layers, introduce low‑voltage micro‑pattern detectors and fast‑fiber timing to achieve ≤100 ps resolution, and continue data taking through Run 4. These upgrades aim to make ANUBIS the most sensitive LLP detector at the LHC, providing complementary coverage to existing experiments and opening new discovery opportunities for hidden sectors.