⚛️ phenomenology

MeV dark matter decay forecasts for the COSI mission

This paper presents forecasts demonstrating that the upcoming COSI mission will improve sensitivity to dark matter decay lifetimes by more than an order of magnitude in the 0.4 to 10 MeV mass range through dedicated all-sky simulations of the Milky Way halo.

Original authors: Robin Anthony-Petersen, Orion Ning, Yu Watanabe, Benjamin R. Safdi, Maíra Dutra, Francesca Calore, Savitri Gallego, William Guan, Christopher M. Karwin, Alexander Kusenko, Israel Martinez-Castellanos

Published 2026-10-14

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Robin Anthony-Petersen, Orion Ning, Yu Watanabe, Benjamin R. Safdi, Maíra Dutra, Francesca Calore, Savitri Gallego, William Guan, Christopher M. Karwin, Alexander Kusenko, Israel Martinez-Castellanos, Shigeki Matsumoto, Michela Negro, Uwe Oberlack, Field R. Rogers, Thomas Siegert, Tadayuki Takahashi, Haoyu Xie, Hiroki Yoneda, Andreas Zoglauer, Steven E. Boggs, Dieter H. Hartmann, Carolyn A. Kierans, John A. Tomsick

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

For decades, astronomers have been searching for the invisible substance that holds galaxies together, a substance known as dark matter. While we cannot see it directly, we know it exists because of how its gravity pulls on stars and gas. The leading theory suggests that dark matter is made of particles, and while these particles are incredibly stable, they might not be perfectly eternal. Some models of physics predict that these particles could eventually decay, breaking apart into lighter particles, including photons, which are particles of light. If this happens, it would create a very specific, faint signal: a single, sharp line of light at a precise energy level, appearing against the background glow of the universe. Finding this signal would be a monumental discovery, revealing the identity of dark matter and opening a window into the fundamental laws of physics that govern our existence.

The challenge has been that this potential signal is expected to appear in a part of the light spectrum known as the "MeV gap," a range of energies between what X-ray telescopes and high-energy gamma-ray detectors can see well. This gap has remained largely unexplored because previous instruments lacked the necessary sensitivity and clarity to spot such a faint, specific line. However, a new space telescope called COSI, scheduled to launch in August 2027, is designed specifically to fill this gap. In a recent study, a team of researchers used advanced computer simulations to forecast what COSI will be able to see once it is in orbit. Their work suggests that this new telescope will be able to probe the decay of dark matter with a sensitivity more than ten times better than any previous attempt, potentially reaching back to timescales of $10^{30}$ seconds for dark matter particles with masses between 0.4 and 10 MeV.

The researchers began by building a detailed digital model of the COSI telescope and the environment it will observe. COSI works by detecting gamma rays through a process called Compton scattering, where a high-energy photon hits an electron inside the detector, bounces off, and loses some energy. By tracking the path and energy loss of these interactions, the telescope can reconstruct the direction and energy of the original light source. The team simulated two years of observations, filling the digital sky with the expected background noise from the universe, such as radiation from distant galaxies and the Earth's atmosphere, as well as the faint, theoretical signal of decaying dark matter. They then tested different ways to analyze this data, looking for that single, sharp line of light that would stand out against the chaotic background.

One of the most significant findings of the study is that the telescope's ability to see the entire sky at once is just as important as its sharpness. Previous estimates of COSI's capabilities relied on looking at just a small, bright patch of the sky near the center of our galaxy. This new analysis, however, accounted for the fact that dark matter is spread out across the entire Milky Way halo. By using the full field of view, the telescope can collect a much larger number of photons over time. The simulations showed that this approach, combined with COSI's low background noise and high energy resolution, would allow it to improve upon current limits by more than an order of magnitude. In simpler terms, if the current best instruments are like trying to hear a whisper in a noisy room, COSI will be like listening in a quiet library, capable of detecting a sound that was previously impossible to hear.

The team also had to account for the reality that the universe is messy and that their models of the background noise might not be perfect. They tested various scenarios where the background was slightly misidentified or where unexpected lines of light appeared. They found that even with these realistic complications, the telescope would still be able to set strict limits on how long dark matter particles can live. The study concludes that if dark matter does decay into two photons within the mass range of 0.4 to 10 MeV, COSI will significantly improve the sensitivity to detect such decay or, if no signal is found, set much tighter constraints on the particle's lifetime than currently possible. This would effectively rule out a wide range of particle physics models that rely on dark matter decaying in this specific energy range within the timescales probed by the mission.

Beyond just finding or ruling out dark matter decay, the study translates these findings into constraints on specific types of hypothetical particles, such as axion-like particles and sterile neutrinos. These are particles that have been proposed to solve other mysteries in physics but have never been directly detected. The simulations show that COSI will be able to test the properties of these particles with a precision that has never been achieved before. For instance, the telescope could probe the strength of the interaction between axion-like particles and light, testing theories that arise from string theory and other high-level physics frameworks. The results suggest that the upcoming mission will not only bridge the gap in our observational capabilities but also probe a meaningful parameter space that naturally arises in top-down string theory constructions.

The path to these results involved a massive computational effort, using powerful supercomputers to simulate the journey of millions of photons through the detector. The researchers had to model the complex geometry of the telescope, the behavior of the detector materials, and the statistical methods needed to separate a faint signal from a noisy background. They developed new ways to handle the data, ensuring that the analysis would remain robust even if the background models were slightly off. This careful preparation means that when COSI finally launches and begins its two-year mission, the scientific community will have a clear, reliable roadmap for interpreting the data. The study does not claim to have found dark matter; rather, it provides a highly confident forecast of what the telescope will be able to do, setting the stage for a potential discovery that could reshape our understanding of the cosmos.

Ultimately, this work represents a shift from speculation to precision in the search for dark matter. By focusing on the MeV gap, a region that has been overlooked for decades, the researchers have identified a promising avenue for discovery. The simulations confirm that the combination of COSI's wide view, its ability to measure energy with great precision, and its low noise levels makes it the ideal tool for this job. If dark matter decays in the way these models predict, the signal will be there, waiting to be found. If it does not, the telescope will tell us that, narrowing the search and forcing physicists to rethink their theories. Either outcome will be a major step forward, turning a long-standing mystery into a concrete piece of scientific knowledge.

Technical Summary: MeV Dark Matter Decay Forecasts for the COSI Mission

Problem Statement
The decay of dark matter (DM) into Standard Model final states is a generic prediction of many particle physics models, particularly those involving axion-like particles, sterile neutrinos, and hidden-sector scenarios. These models often predict distinctive monochromatic signatures (e.g., $\chi \to \gamma\gamma$ ) in the "MeV gap" ( $\sim 0.1$ –$100$ MeV). However, this energy range has remained under-explored compared to the keV and GeV regimes. Existing constraints in the MeV band rely on decades-old instruments, specifically INTEGRAL/SPI and COMPTEL, which are limited by lower sensitivity and higher systematic uncertainties. There is a critical need for a new instrument capable of probing DM lifetimes in this mass range with high spectral resolution and low background to test these well-motivated particle physics scenarios.

Methodology
The authors present projected sensitivity forecasts for the upcoming Compton Spectrometer and Imager (COSI) mission, a NASA Small Explorer scheduled for launch in August 2027. The study utilizes a comprehensive forward-modeling framework rather than relying on simplified point-source sensitivity estimates used in previous projections.

Signal and Background Modeling:
- Signal: The DM signal is modeled as a monochromatic photon line ( $E_\gamma = m_\chi/2$ ) originating from the Galactic halo, assuming a Navarro–Frenk–White (NFW) density profile. The signal is propagated through a detailed GEANT4-based model of the COSI instrument using the MEGAlib library.
- Background: The analysis incorporates the fourth COSI Data Challenge (DC4) mock datasets, which simulate three months of observation. These include astrophysical backgrounds (Galactic diffuse continuum, extragalactic background, specific sources like $^{26}$ Al and $^{60}$ Fe) and instrumental backgrounds (activation lines, albedo protons, etc.).
- Exposure Scaling: Results from the 3-month simulation are scaled to a nominal 2-year mission lifetime, assuming the background level remains constant for the continuum (though noting that instrumental activation lines may increase).
Analysis Frameworks:
The paper evaluates two primary analysis approaches to assess sensitivity to the DM lifetime ( $\tau_\chi$ ):
- Energy-Only Analysis: This approach sums events over the angular coordinates of the Compton Data Space (CDS) to create a 1D energy spectrum. It employs a sliding energy window ( $\pm 2\sigma_E$ $\pm 2 σ_{E}$ ) around the target mass. To address systematic mismodeling, the authors test several variants:
  - Optimal: Assumes a perfect background model.
  - Sliding Window Power-Law: Fits a power-law continuum with nuisance parameters for normalization and spectral index.
  - Bright-Line Masking: Masks energy bins around bright astrophysical/instrumental lines ( $\sigma_l \ge 30$ ).
  - Dim-Line Modeling: Includes nuisance parameters for dim lines ( $\sigma_l < 30$ ).
  - Systematic Mitigation: Introduces a "spurious signal" nuisance parameter (following ATLAS and X-ray line search methodologies) and treats the uncertainty $\sigma_i$ as a free parameter to ensure the distribution of test statistics (TS) aligns with the null hypothesis.
- Compton Data Space (CDS) Analysis: This utilizes the full 4D data space (energy $E$ , scattering angle $\phi$ , and Galactic coordinates $\ell, b$ ). Because the background realization is sparse in 4D, the authors apply background smoothing across energy bins to enable statistical analysis. This represents a "best-case" scenario where background morphology is perfectly known, isolating the potential gain from spatial information.

Key Contributions

Full Instrument Response Modeling: Unlike previous works (e.g., Ref. [32]) that extrapolated from point-source sensitivities, this study forward-models the extended DM signal through the full COSI instrument response, including detailed event reconstruction and selection cuts.
Systematic Uncertainty Quantification: The paper rigorously tests the impact of background mismodeling. It demonstrates that while a "perfect" background model yields the best limits, realistic analysis frameworks incorporating nuisance parameters for continuum shape, line masking, and spurious signals maintain robust sensitivity, albeit with some degradation.
Comprehensive Sky Coverage: The analysis utilizes the entire sky, leveraging COSI's large field of view ( $>25\%$ of the sky) rather than restricting the search to the Galactic Center, significantly increasing the effective exposure.

Results

Sensitivity Improvement: COSI is projected to improve the sensitivity to the DM decay lifetime by more than an order of magnitude over existing constraints from INTEGRAL and COMPTEL across the mass range of $0.4$ to $10$ MeV.
Lifetime Limits: For a 2-year mission, COSI is expected to probe DM lifetimes up to $\tau_\chi \sim \mathcal{O}(10^{30})$ s for masses below $\sim 4$ MeV. Sensitivity weakens at higher masses due to decreased effective area from incomplete containment of high-energy $\gamma$ -rays.
Analysis Comparison:
- The "optimal" energy-only analysis (perfect background) sets the upper bound on sensitivity.
- Realistic sliding-window analyses degrade sensitivity by a factor of $\sim 2$ or more compared to the optimal case, with degradation increasing at higher masses.
- The inclusion of the "spurious signal" nuisance parameter successfully aligns the distribution of discovery test statistics with the null hypothesis expectation, correcting for mismodeling at the cost of slightly reduced sensitivity at specific masses.
- The full CDS analysis provides only a marginal improvement (median factor of $\sim 1.15$ ) over the optimal energy-only analysis. This is attributed to the fact that the diffuse NFW halo signal is not sharply distinguished from the diffuse background by morphology alone.
Particle Physics Implications: The projected limits are recast into constraints on specific models:
- Axion-Like Particles (ALPs): COSI will probe axion-photon couplings ( $g_{a\gamma\gamma}$ ) down to $\sim 10^{-20}$ GeV $^{-1}$ , covering parameter space naturally predicted by string theory compactifications.
- Sterile Neutrinos: The sensitivity reaches mixing angles $\sin^2(2\theta)$ at the level of $10^{-22}$ for MeV-scale sterile neutrinos, probing production mechanisms independent of active-sterile oscillations.

Significance and Claims
The paper concludes that the COSI mission represents a major step forward for MeV astrophysics and beyond-the-Standard-Model physics. By bridging the "MeV gap" with high spectral resolution and low background, COSI will significantly enhance the ability to search for decaying dark matter. The authors emphasize that their projections, which account for realistic systematic uncertainties and full instrument response, are more accurate and optimistic than previous estimates that relied on point-source scaling. The mission offers the potential to discover the identity of dark matter or place stringent constraints on particle physics models predicting MeV-scale decays. The authors note that while the current study focuses on monochromatic lines, future work will assess sensitivity to continuum signatures from DM annihilation or decay into unstable particles.

Technical Summary: MeV Dark Matter Decay Forecasts for the COSI Mission

More like this