Electron signals could extend LZ’s search for lightweight dark matter

Dark matter searches face a practical problem at low particle masses: a collision can leave too little energy for a detector to register. A new preprint explores how electrons shaken loose during such a collision could provide a more visible signature in the LUX-ZEPLIN experiment, known as LZ.

Yang Yu and Lei Feng, affiliated with China’s Purple Mountain Observatory and the University of Science and Technology of China, estimate sensitivity to dark matter masses from about 350 megaelectronvolts to 10 gigaelectronvolts, expressed in particle-physics mass units. Their work was submitted to arXiv on November 6, 2026; the supplied record does not establish peer review.

Crucially, this is an expected-sensitivity calculation using background-only simulated counts and an exposure drawn from published LZ results. It is neither a dark matter detection nor a newly reported experimental exclusion from the LZ collaboration.

A second signature from a small collision

Conventional direct-detection searches look for a nucleus recoiling after a hypothetical dark matter particle strikes it. When the incoming particle is much lighter than the nucleus, the maximum recoil energy drops rapidly with decreasing particle mass.

The Migdal effect offers another route. If a nucleus suddenly recoils, its surrounding electrons may not adjust immediately. An electron can be excited or ejected, and the atom’s subsequent rearrangement can release additional electrons or X-rays. These electronic energy deposits can be detectable even when the nuclear recoil alone falls below the usual threshold. The process redistributes energy from the collision; it does not create extra energy.

Yu and Feng calculated the expected electronic signal in xenon by combining nuclear-scattering rates with published atomic ionization probabilities. They included all occupied xenon electron levels in the calculation, although only energetically accessible shells contribute to a given signal. Detector charge yield and detection efficiency then translate the prediction into an observable spectrum.

What the projected limits mean

The researchers generated signal spectra at 20 masses between 100 megaelectronvolts and 10 gigaelectronvolts. They assumed a standard model for dark matter’s Galactic speed distribution and examined electronic energy deposits between 1 and 4 kiloelectronvolts.

Their calculation uses 4.2 tonne-years of exposure—a measure combining detector target mass and operating time. Instead of fitting an observed event sample, they used an “Asimov” dataset: artificial counts set exactly equal to the expected background. This estimates sensitivity without the random fluctuations present in actual measurements.

A Poisson statistical model described the counts in each energy bin. The fit allowed the overall background and signal strengths to vary, included a 20% signal-normalization uncertainty, and assigned 30% relative uncertainty to individual background-bin rates. Sampling the resulting probability distribution yielded an upper limit containing 90% of the marginal posterior probability for signal strength. These are Bayesian limits, not frequentist confidence intervals.

Across the accessible mass range, the authors report projected upper limits on the spin-independent dark matter–nucleon scattering cross section ranging from 6.5 × 10⁻³⁷ to 9.6 × 10⁻⁴² square centimetres. Cross section measures interaction strength; spin-independent means the leading interaction does not depend on particle spin orientation.

Below roughly 350 megaelectronvolts, their assumed dark matter population cannot produce events in the selected energy window, so the calculation gives no finite limit. Just above that threshold, only the fastest particles contribute. Above about 1 gigaelectronvolt, the authors find their projected LZ limit stronger than the Migdal limits from the other experiments they compare—not necessarily stronger than every dark matter constraint.

Model dependence matters

The study also translates its projections into constraints on hypothetical force-carrying particles, called mediators, connecting dark matter to ordinary matter. For a representative dark matter mass of 1 gigaelectronvolt, the authors retain the scattering’s momentum dependence and consider scalar and vector mediators, distinguished by their spin.

At low mediator masses, they obtain a bound around 10⁻¹² on a dimensionless combination of the mediator’s couplings to dark matter and nucleons. They report competitive or stronger sensitivity than selected bounds from kaon decays, stellar cooling and supernova SN1987A. Those comparisons require additional, consistent assumptions about mediator production, decay and interactions; they are not universal rankings.

The projected reach also depends on atomic ionization calculations, liquid-xenon response, energy calibration, efficiency and background modeling. The paper notes that changing dark matter’s assumed speed distribution can shift low-mass limits by roughly 10–100%. The result therefore identifies a promising search channel under specified assumptions, rather than establishing that lightweight dark matter has been found or definitively ruled out.