⚛️ phenomenology

Constraints on Axion-like Dark Matter from Cosmic Birefringence with a Polarization Array of Repeating Fast Radio Bursts

This paper establishes a Bayesian framework to search for axion-like dark matter using polarization data from hyperactive repeating fast radio bursts, finding no evidence for such dark matter and setting stringent new upper limits on the Chern-Simons coupling across a broad mass range.

Original authors: Xiaohui Liu, Zi-Yan Yuwen, Yun-Long Zhang, Zerui Liu, Shiqian Zhao, Shuai Feng, Wei-Yang Wang, Xuelei Chen

Published 2026-09-22

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Xiaohui Liu, Zi-Yan Yuwen, Yun-Long Zhang, Zerui Liu, Shiqian Zhao, Shuai Feng, Wei-Yang Wang, Xuelei Chen

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

The universe is filled with invisible stuff. We know it is there because its gravity holds galaxies together and shapes the cosmos, yet we cannot see it, touch it, or detect it with our standard instruments. This mysterious substance is called dark matter. For decades, scientists have proposed many ideas about what it might be, ranging from heavy, slow-moving particles to something far stranger: a field of ultra-light particles that behave like waves stretching across the entire sky. One leading candidate for this wave-like dark matter is the axion, a particle so light that a single one would be billions of times lighter than an electron. If these particles exist, they might interact with light in a very specific way, twisting the direction in which light waves vibrate as they travel through space. This twisting effect, known as cosmic birefringence, would leave a faint, rhythmic signature on the light coming from distant objects, acting like a cosmic fingerprint for dark matter.

A team of researchers has now turned their attention to some of the most energetic and distant beacons in the universe to hunt for this signature: fast radio bursts. These are intense flashes of radio waves that last only a fraction of a second, originating from galaxies billions of light-years away. Some of these bursts repeat, firing off multiple times from the same source, and many of them are highly polarized, meaning their light waves vibrate in a specific, organized direction. The researchers focused on three particularly active repeating sources, using the Five-hundred-meter Aperture Spherical radio Telescope, known as FAST, in China to monitor them over long periods. By tracking the precise angle of the light's vibration from these bursts over time, the team looked for the subtle, rhythmic oscillations that would indicate the presence of axion-like dark matter.

The scientists developed a sophisticated method to analyze the data, treating the collection of radio bursts as a single, powerful network. They had to account for the fact that the light from these bursts naturally jitters in its polarization angle, a bit like static on a radio line, which could easily hide the signal they were looking for. To solve this, they built a statistical model that could separate the natural noise of the bursts from the potential signal of dark matter. They also calculated how much dark matter should be present at the location of each burst's host galaxy, using computer simulations of galaxy formation to estimate the density of the invisible matter surrounding these distant worlds. This allowed them to predict exactly what the dark matter signal should look like if it were there, comparing the prediction against the actual observations from Earth.

After analyzing years of data from the three repeating sources, the team found no evidence of the rhythmic twisting that axion-like dark matter would produce. The polarization angles of the radio bursts remained steady, showing no signs of the oscillating background that the theory predicted. While this result does not prove that axion-like dark matter does not exist, it does rule out a wide range of possibilities for how strongly these particles might interact with light. The researchers set strict new limits on the strength of this interaction, effectively narrowing the search space for these elusive particles. Their findings suggest that if axion-like dark matter is present, it is not interacting with light strongly enough to be detected by these specific bursts over the time periods observed.

The study also demonstrated the power of combining data from multiple sources. By treating the three different fast radio bursts as a single array, the team was able to improve their sensitivity, particularly for the lightest and slowest-oscillating dark matter candidates. This approach allowed them to look for signals over longer time scales than any single source could provide on its own. Although the current data did not reveal a discovery, the method proved to be a robust and promising tool for future searches. As more repeating fast radio bursts are discovered and monitored for longer periods, this technique will become even more powerful, offering a new and independent way to probe the nature of the dark matter that permeates our universe. The search continues, but the path forward is now clearer, guided by the precise measurements of light from the deep cosmos.

Technical Summary: Constraints on Axion-like Dark Matter from Cosmic Birefringence with a Polarization Array of Repeating Fast Radio Bursts

Problem Statement
The nature of dark matter (DM) remains a central puzzle in modern physics. Axion-like dark matter (ALDM) is a well-motivated candidate that couples to photons via a Chern-Simons interaction. This coupling induces cosmic birefringence, causing the polarization angle (PA) of linearly polarized electromagnetic waves to oscillate temporally as they propagate through the ALDM field. While previous studies have utilized pulsars and other astrophysical sources to search for these oscillations, the field requires probes with higher flux densities and distinct spatial characteristics to improve constraints. Repeating Fast Radio Bursts (FRBs) offer a promising avenue due to their extreme brightness and high linear polarization, yet their application is complicated by substantial intrinsic burst-to-burst PA scatter and the need to model the DM density at cosmological distances.

Methodology
The authors develop a hierarchical Bayesian framework to search for ALDM-induced PA oscillations in repeating FRBs. The methodology is structured as follows:

Theoretical Model: The ALDM-induced PA shift, $\Delta\psi_a$ , is modeled as the difference between the ALDM field at the source and the Earth. The field is treated as a stochastic superposition of harmonic oscillations. Crucially, the model incorporates a two-component birefringence approach: the source term (oscillating at a redshifted frequency $\omega_s = m_a/(1+z)$ ) and the Earth term (oscillating at $\omega_0 = m_a$ ). The authors argue that because the inferred DM densities at the FRB host locations are comparable to the local Galactic DM density, neither term can be neglected.
Inference Framework:
- Individual Sources: For single repeating FRBs, the framework treats the ALDM signal as a deterministic realization of the source and Earth fields. It simultaneously infers the axion parameters and the population-level properties, specifically the intrinsic burst-to-burst PA scatter ( $\sigma$ ) and the mean PA ( $\mu$ ).
- FRB Polarization Array (FRBPA): To combine multiple sources, the authors construct an FRBPA. Unlike pulsar arrays where sources and Earth share a common halo, FRB sources reside in distinct halos. Consequently, the source terms are statistically independent, and correlations arise solely from the common Earth term. The ALDM signal is marginalized analytically and encoded into an additional covariance matrix ( $C_a$ ) that includes off-diagonal elements representing the Earth-term correlation between different FRBs.
Dark Matter Density Estimation: To determine the source DM density ( $\rho_s$ ), the authors utilize the UNIVERSEMACHINE empirical galaxy-halo connection model. They match the observed stellar mass and star formation rate (SFR) of FRB host galaxies to simulated galaxy catalogs. Using a simulation-calibrated generalized Navarro-Frenk-White (gNFW) profile, they derive the local DM density at the specific projected offset of each FRB within its host halo.
Data Processing: The analysis uses PA datasets from three hyperactive repeating FRBs (FRB 20201124A, FRB 20220912A, and FRB 20240114A) observed by the Five-hundred-meter Aperture Spherical radio Telescope (FAST). The authors apply strict quality cuts, retaining only bursts with high signal-to-noise ratios (SNR > 50) and small PA excursions to minimize intrinsic model mismatch.

Key Contributions

Hierarchical Bayesian Framework: The paper introduces a novel statistical approach that explicitly models intrinsic PA scatter while searching for ALDM signals, allowing for robust constraints even in the presence of significant source variability.
FRB Polarization Array (FRBPA): This work presents the first application of an FRBPA concept. It demonstrates how to combine multiple repeating FRBs by exploiting the temporal correlation of the common Earth term while treating source terms as independent.
Simulation-Calibrated Density Estimation: Instead of relying on single-galaxy proxies, the authors employ a population-based approach using cosmological simulations to estimate DM densities at FRB host locations, providing a statistically representative and conservative estimate of the source contribution to the birefringence signal.
Two-Component Model: The authors justify and implement a full two-component model (source + Earth) for the birefringence signal, moving beyond approximations that assume dominance by a single term.

Results

No Detection: The analysis finds no statistically significant evidence for ALDM-induced PA oscillations in either the individual source analyses or the joint FRBPA analysis. The Bayes factors (BF) comparing the ALDM model to the null model remain below the threshold for "negligible evidence" (BF < 100.0) across the mass range, with most values falling below unity, indicating the data generally favor the null hypothesis (where BF < 1 implies the ALDM scenario is disfavored).
Upper Limits: The authors place 95% credible upper limits on the Chern-Simons coupling strength ( $g_{a\gamma\gamma}$ $g_{aγ γ}$ ) as a function of ALDM mass ( $m_a$ $m_{a}$ ).
- The constraints cover the mass range $10^{-23}$ to $10^{-18}$ eV.
- The resulting upper limits on $g_{a\gamma\gamma}$ for the FRBPA lie between $10^{-12.6}$ and $10^{-7.9}$ GeV $^{-1}$ .
- The FRBPA analysis improves constraints at the low-mass end ( $10^{-23} \lesssim m_a \lesssim 10^{-22}$ eV) by a factor of 2–10 compared to individual source limits, primarily due to the combined observing spans and phase coverage.
Comparison: The constraints are comparable to the upper limit from the Simons Observatory, although they remain weaker than several other astrophysical limits (e.g., CMB polarization, pulsar arrays, gamma-rays) shown in the study.

Significance and Claims
The paper positions long-term polarimetric monitoring of hyperactive repeating FRBs as a "promising and independent probe" of ALDM. The authors modestly characterize the current results as a "proof-of-concept demonstration" rather than the ultimate constraint, noting that the longest individual-source baseline is approximately 472 days. They emphasize that the sensitivity of the method is governed by the intrinsic variance of the PA distribution, the number of detected bursts, and the temporal coverage of the axion-induced phase evolution.

The study concludes that as monitoring campaigns continue and larger samples of repeating FRBs accumulate over longer time baselines—particularly with future facilities like CHORD and the SKA—the sensitivity to low-mass ALDM is expected to improve substantially. The work establishes the viability of the FRBPA method for future searches, highlighting its ability to mitigate the limitations of individual source activity patterns through joint analysis.

More like this