⚛️ phenomenology

Boosted Dark Matter Directionality in Large Liquid Scintillator Detectors

This paper proposes that large liquid scintillator detectors can probe new parameter space for boosted light dark matter by utilizing neutron-coincidence signals from quasi-elastic scattering off carbon, though the potential for directional reconstruction is limited by large nucleon scattering angles.

Original authors: Samuel S. H. Tse, Qishan Liu, Kenny C. Y. Ng, Koun Choi, Yu-Feng Li

Published 2026-10-23

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Samuel S. H. Tse, Qishan Liu, Kenny C. Y. Ng, Koun Choi, Yu-Feng Li

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Dark matter is the invisible substance that holds galaxies together, yet despite decades of searching, no one has ever directly caught a single particle. We know it exists because of how stars move and how light bends around massive clusters, but its true nature remains a mystery. In standard searches, scientists look for tiny bumps caused when a slow-moving dark matter particle hits an atom in a detector deep underground. However, if the dark matter is very light, these bumps are too faint to see with current technology. A newer idea suggests that dark matter might sometimes get a "boost," gaining tremendous speed by colliding with high-energy cosmic rays traveling through space. If this happens, the dark matter arrives at Earth with enough energy to create a detectable signal, turning a ghostly whisper into a shout that large detectors might finally hear.

A team of researchers recently investigated how to catch these fast-moving particles using massive tanks of liquid scintillator, a type of clear, oily fluid used in neutrino experiments. They focused on a specific interaction where a boosted dark matter particle smashes into a carbon atom inside the detector and knocks a neutron out of the nucleus. Unlike the charged particles that usually light up these detectors, a neutron is electrically neutral and travels invisibly for a short time before it gets caught by another atom. When this capture happens, it releases a distinct burst of light roughly 210 microseconds later. This creates a unique signature: a prompt flash of light from the initial collision, followed by a delayed flash from the neutron capture. By linking these two events, scientists can confirm that a neutron was indeed knocked out, filtering out much of the background noise that plagues other searches.

The researchers simulated this process in a hypothetical detector the size of 100,000 tons of liquid, a scale comparable to the largest neutrino observatories currently in operation. They calculated how many of these boosted dark matter particles would reach the detector after traveling through the Earth, accounting for the fact that the planet itself acts as a shield, slowing some of them down or stopping them entirely. Their simulations showed that this "neutron-knockout" channel is a powerful tool. Because carbon atoms contain six neutrons each, while hydrogen atoms (the other common target in these detectors) have none, the liquid scintillator offers a much larger target for this specific type of interaction. The study found that with ten years of data from such a detector, scientists could probe regions of dark matter properties that were previously inaccessible, offering tighter constraints than those found by looking for signals from free protons alone.

A central question the team asked was whether they could use the direction of these events to improve their search. Since the source of boosted dark matter is expected to be the center of our galaxy, the particles should arrive from a specific direction in the sky. If the detector could tell which way the neutron was traveling, it could distinguish the signal from random background events that come from all directions. To test this, the team modeled how the neutron moves after being knocked out. They found that while the neutron does travel in a general direction, it scatters wildly as it moves through the liquid, bouncing off other atoms and losing its original path. By the time it is captured and the second flash of light is recorded, the connection to its original direction is heavily blurred.

The results of the simulation were clear: while the detector can successfully identify the neutron events, the directional information does not significantly improve the ability to find dark matter. The scattering angle of the neutron is so large that it washes out the directional signal, making the map of events look almost the same whether the scientists try to use the direction or ignore it. The researchers concluded that the only way to gain a real advantage from directionality would be to find a process where the particles travel in a much straighter line, or to look for dark matter sources that are even more concentrated than the galactic center. For now, the best strategy remains using the sheer volume of the detector and the unique timing of the neutron capture to sift through the noise. This work provides a realistic roadmap for future experiments, showing that while the dream of a directional dark matter detector is compelling, the physics of neutron scattering sets a firm limit on how much the direction can help.

Technical Summary: Boosted Dark Matter Directionality in Large Liquid Scintillator Detectors

Problem Statement
Conventional direct detection of sub-GeV dark matter (DM) is kinematically limited, as halo DM particles typically possess velocities ( $v \sim 10^{-3}c$ ) insufficient to generate nuclear recoil energies above detector thresholds. However, if DM is boosted to relativistic energies via interactions with cosmic rays (Cosmic-Ray-Boosted Dark Matter, or CRBDM), it can trigger detectable signals in large detectors. While previous studies have utilized CRBDM scattering off free protons in liquid scintillator detectors (LLSDs), this work investigates the potential of scattering off bound neutrons in carbon nuclei. Specifically, the authors examine whether the directional information preserved in the displacement between the neutron emission vertex and its capture vertex can significantly improve the sensitivity to spin-independent DM–nucleon scattering cross-sections ( $\bar{\sigma}_n$ ) compared to non-directional analyses.

Methodology
The study employs a multi-stage theoretical and simulation framework:

Flux Calculation: The authors model the CRBDM flux arriving at Earth using a cylindrical leaky-box model for cosmic rays and a Navarro–Frenk–White (NFW) profile for the Galactic DM density. They calculate the differential flux of boosted DM particles ( $\chi$ ) resulting from cosmic ray (proton and helium) scattering, accounting for Earth attenuation effects which shift the energy spectrum and reduce flux depending on the zenith angle.
Interaction Modeling: The signal process is defined as the quasi-elastic scattering (QES) of boosted DM off a bound neutron in a $^{12}\text{C}$ nucleus ( $\chi + ^{12}\text{C} \to \chi + n + ^{11}\text{C}^*$ ). The differential cross-section is calculated using the Impulse Approximation (IA) and a Relativistic Fermi Gas (RFG) model, incorporating Pauli blocking and final-state interactions (FSI) via a nuclear optical potential.
Neutron Propagation and Vertex Reconstruction: Using Geant4 (v11.2.2), the authors simulate the propagation of knocked-out neutrons in a liquid scintillator (modeled after JUNO's composition). They track:
- Prompt Signal: The secondary scintillation light generated by the neutron's inelastic scattering and the de-excitation of the residual nucleus.
- Delayed Signal: The capture of the thermalized neutron (primarily on Hydrogen, emitting a 2.2 MeV gamma ray) after a mean time of $\sim 210\,\mu\text{s}$ .
- Directionality: The displacement vector between the reconstructed prompt vertex and the delayed capture vertex. The simulation accounts for vertex reconstruction uncertainties ( $R_{\text{int}}$ and $R_{\text{cap}}$ ) and neutron diffusion.
Background and Sensitivity Analysis: The primary background is identified as atmospheric neutrino neutral-current quasi-elastic scattering (NCQES) on carbon, which produces an identical signature (prompt + delayed neutron capture). The authors construct a joint likelihood function across 768 sky pixels and 15 energy bins to derive 95% confidence level (CL) limits on $\bar{\sigma}_n$ , comparing scenarios with and without directional information.

Key Contributions

Neutron-Knockout Channel: The paper establishes the feasibility of using the neutron-knockout channel in LLSDs for CRBDM searches. This channel offers a higher target density (bound neutrons in carbon are $\sim 3.6\times$ more abundant than free protons) and a lower background rate due to the prompt-delayed coincidence requirement, which suppresses single-energy deposition backgrounds.
Directional Sensitivity Quantification: The study rigorously tests the hypothesis that the displacement between the neutron emission and capture vertices can serve as a directional tag. It quantifies the "smearing" effect caused by the large nucleon scattering angle and neutron diffusion.
Comparative Limits: The authors provide updated sensitivity projections for a 100 kt $\cdot$ yr exposure, comparing the neutron channel against previous free-proton channel results and other next-generation experiments (JUNO, Hyper-K, DUNE).

Results

Sensitivity Improvement: The inclusion of directional information yields only a marginal improvement in sensitivity. For a mediator mass $m_\phi = 3$ GeV and DM mass $m_\chi = 1$ MeV, the improvement in the cross-section limit is approximately 3% compared to the non-directional case.
Cause of Limited Gain: The authors attribute this lack of significant improvement to the intrinsic physics of the scattering process. The large scattering angle of the knocked-out neutron relative to the incoming BDM direction, combined with neutron diffusion and vertex reconstruction uncertainties, effectively washes out the directional correlation. Even in an idealized scenario where diffusion is removed, the scattering angle alone prevents a sharp concentration of events toward the Galactic Center.
Channel Comparison: The neutron-knockout channel provides stronger constraints than the free-proton channel (Ref. [1]) for the same exposure, primarily due to the larger target mass and superior background rejection. At $m_\phi = 1$ GeV, the projected sensitivity reaches the level expected for Hyper-K with 5 years of data, though DUNE remains the most sensitive projected experiment in this framework.
Sky Map Smearing: Simulations show that while the unattenuated BDM flux is highly concentrated toward the Galactic Center, the reconstructed neutron emission map is significantly smeared. The ratio of the brightest to dimmest sky pixels drops from $\sim 83$ (unattenuated BDM) to $\sim 8$ (after nucleon scattering) and further to $\sim 2.5$ (after diffusion and vertex reconstruction).

Significance and Claims
The paper concludes that while the neutron-knockout channel in LLSDs is a powerful tool for probing CRBDM parameter space—offering better sensitivity than free-proton searches—the utility of directional information in this specific channel is limited. The inherent directionality of CRBDM is insufficient to overcome the smearing effects of the scattering kinematics.

The authors modestly claim that the primary value of utilizing directionality in this context is that it comes at "essentially no cost," relying on vertex reconstruction techniques already available in LLSDs. They suggest that future improvements in directional sensitivity would require channels with more sharply concentrated fluxes, such as DM annihilation into BDM or neutrinos (where the J-factor scales as $\rho_\chi^2$ ), rather than relying on the scattering kinematics of CRBDM off nucleons. The work encourages LLSD experiments to continue exploring directionality but cautions against expecting dramatic gains from the CRBDM mechanism alone.

More like this