The search for dark matter in direct detectors has gone two decades without an unambiguous signal, and sensitivity is approaching the irreducible neutrino background. Against that backdrop, a single high-energy candidate at 248 keV, recorded by LZ with an exposure of 2.84 tonne-years, looks intriguing: the background hypothesis is rejected at the 2.6σ level. The problem is not the recoil energy itself, but that elastic scattering of the dark matter halo would produce a monotonically falling spectrum with a noticeable excess at low energies, which is not observed. To transfer a momentum of about 246 MeV to a xenon nucleus, you need either a heavy particle with a mass greater than 79 GeV or a light but relativistic one. The work stands in a line of numerous attempts to explain this event, but it differs in systematically comparing two mechanisms for suppressing the low-energy tail within a single model-independent framework.
The authors use effective field theory with contact operators of dimension six: scalar-scalar, pseudoscalar-scalar, and pseudoscalar-pseudoscalar. For inelastic dark matter, a mass splitting δ between two states is introduced, which kinematically forbids low-energy recoils. For boosted dark matter, a light particle is considered, produced when a heavier component annihilates in the halo, which yields a monoenergetic flux with a Lorentz factor equal to the ratio of the masses. For each combination, a full predicted spectrum is constructed taking into account the Helm nuclear form factor, the LZ detection efficiency, and the energy resolution, and then a likelihood profile is performed over the interaction scale Λ. Comparison with the single event reduces to the shape of the normalized spectrum, since Λ drops out of the shape analysis.
The two scenarios give qualitatively different spectra. For inelastic dark matter with a mass of 1 TeV and a splitting of 300 keV, the fraction of events below 150 keV is 17.4% for the scalar operator, 14.1% for pseudoscalar-scalar, and 6.0% for pseudoscalar-pseudoscalar. For boosted dark matter with a mass of 10 GeV and a momentum of 500 MeV, the same fractions are 99%, 92%, and 25%, respectively. That is, in the boosted scenario only the pseudoscalar-pseudoscalar interaction places most events above 150 keV, which agrees with the observation. In the inelastic case, the preferred splitting δ decreases monotonically as the momentum dependence strengthens: from the scalar operator to the pseudoscalar-pseudoscalar one. The kinematic boundary for the inelastic scenario gives δ ≤ 385 keV, while for the boosted one a mass greater than 78.7 GeV is required if the speed is limited by the Galactic escape velocity.
The work shows that a single event is not enough to choose between inelastic and boosted dark matter, but the shape of the spectrum in the high-energy window is sensitive to the type of interaction. Unlike neighboring papers, which usually consider one specific mechanism, here both scenarios are analyzed on an equal footing within a single effective theory. This makes it possible to compare predictions directly and reveals a degeneracy between the operator structure and the mass splitting in the inelastic case. Such an approach sets a template for future interpretations of rare events in direct-search detectors.
If LZ accumulates its full exposure and records a few more events in the 200-270 keV region, the shape of the spectrum will make it possible to distinguish the inelastic scenario from the boosted one. For boosted dark matter, the key measurement will be the fraction of events below 150 keV: the scalar and pseudoscalar-scalar operators give more than 90% of events there, while the pseudoscalar-pseudoscalar one gives only 25%. Additional constraints could come from detectors with a different target, such as argon or fluorine, where the kinematic thresholds are different. It is also possible to look for signals from the same mechanism in neutrino telescopes or in gamma radiation from the Galactic center.
The results will affect the interpretation of data from all xenon detectors, such as XENONnT and PandaX-4T, as well as the planning of future experiments with larger exposures. They are also useful for theorists building models of a dark sector with multiple states or with light boosted particles.
The immediate step is to analyze the full LZ exposure and search for additional events in the 200-270 keV window. In parallel, it would be worth performing a similar comparison for other effective-theory operators, including spin-dependent ones, and for other halo velocity distribution profiles.
In dark matter physics, the question of the nature of the particle and its interactions remains unresolved. This work is directly connected to the problem of interpreting single events under low-statistics conditions and to the degeneracy between mass, splitting, and interaction structure. It also touches on the question of how much trust can be placed in effective operators at large momentum transfers, when the scale Λ becomes comparable to the momentum.
For boosted dark matter with a pseudoscalar-pseudoscalar interaction, 22% of events fall into the 225-271 keV window, where the recorded candidate lies, whereas for the scalar operator only 0.14% do.
The idea of boosted dark matter resembles plots in which invisible particles gain energy from the decay or annihilation of heavier counterparts, as in some stories about dark sectors, but there is no direct reference to a known work of science fiction.