UCL Reanalysis Confirms 20 GeV Gamma‑Ray Halo Around Milky Way, but Stops Short of Dark‑Matter Claim

A new University College London reanalysis says a previously reported gamma-ray “halo” around the Milky Way still shows up when the data are tested with a more detailed, pixel-by-pixel method, reinforcing the case that the residual is not simply an artifact of a coarse analysis. But the authors are careful not to call it a dark-matter detection.

The study, posted July 9 on arXiv as a preprint, revisits a controversial 2025 claim by astrophysicist Tomonori Totani that Fermi-LAT space telescope data contain a roughly spherical excess of gamma rays peaking near 20 gigaelectronvolts, or GeV, at high latitudes around the galaxy. That earlier result drew attention because such a spectrum could be broadly compatible with annihilating dark matter. It also drew skepticism, because extraordinary claims in gamma-ray astronomy often hinge on how backgrounds are modeled.

That is why the new paper matters. Independent reproduction is a high bar in a field where faint residuals can shift with analysis choices. In “The 20 GeV Galactic Halo Excess: Pixel-Level Confirmation and Consistency with Sub-TeV WIMP Annihilation,” UCL physicists T. R. Stenhouse, C. Ghag and F. F. Deppisch say the signal survives a more granular reanalysis of the same basic question, strengthening the empirical case that there is a real halo-like residual even if its cause remains unsettled.

The team used Fermi-LAT Pass-8 data collected from Aug. 4, 2008, to March 11, 2026. They first recreated Totani’s broader cell-based approach, then extended it with a pixel-level likelihood analysis on native 0.125-degree maps. The update adds energy-dependent forward folding of the instrument’s point-spread function — a way of accounting for how the telescope blurs sources differently at different energies — and masks bright gamma-ray sources that might otherwise contaminate the measurement.

Both approaches recovered a halo-like residual with a spectrum that peaks near 20 GeV, the paper says. The pixel-level fit found a normalization about 20% higher than the cellwise result across the halo scalings the authors tested. They describe the feature as a high-latitude signal that is centrally concentrated around the Milky Way, which they argue makes a simple extragalactic explanation less persuasive. They also stress that this putative halo is distinct from the long-debated GeV excess toward the Galactic center.

The authors then asked whether the residual could be fit by standard weakly interacting massive particle, or WIMP, models in which dark matter annihilates directly into ordinary particles. Their best fits point to masses of about 0.55 teraelectronvolts for the W+W− channel and 0.72 teraelectronvolts for the b quark-antiquark channel, with an annihilation cross section of about 1 × 10^-24 cubic centimeters per second.

That interpretation, however, runs into a major problem. The same rate is in roughly four-to-five-times tension with published limits from dwarf spheroidal galaxies, small satellite galaxies of the Milky Way that have little conventional gamma-ray background and are therefore among the cleanest places to test dark-matter signals. After folding in uncertainties from foreground modeling and from the dark-matter content of those dwarf galaxies, the UCL team broadens the tension to a range of about 1.6 to 9.3 and argues that the simple s-wave annihilation scenario is not strictly ruled out by current systematics.

Totani’s 2025 study, later published in JCAP, had similarly reported a spherical excess peaking near 20 GeV and suggested consistency with dark matter in the roughly 0.5 to 0.8 TeV range. As Totani wrote at the time, “A statistically significant halo-like excess is found with a spectral peak around 20 GeV, while its flux is consistent with zero below 2 GeV and above 200 GeV.”

The new paper briefly considers other explanations, but none offers an easy way out. P-wave annihilation misses relic-abundance requirements by orders of magnitude, dark-matter decay can sidestep dwarf-galaxy limits but conflicts with the isotropic gamma-ray background, and low-velocity enhancement mechanisms could work only with a fine-tuned resonance, the authors say.

Their main caution is explicit. “Throughout, we separate the empirical statement that a 20 GeV halo-like residual is present and robust from the model-dependent claim that it is dark matter, and we stress that our results establish consistency with a dark matter interpretation rather than a detection.”

That distinction is the key takeaway. The UCL reanalysis makes the underlying residual harder to dismiss as a methodological fluke, but it does not resolve what is producing it. The study is an arXiv preprint, meaning it has not yet been peer reviewed. The authors have also released a public GitHub repository for the analysis, a step that should make the result easier for other researchers to check.