⚛️ high-energy experiments

First broadband search for fixed-polarization dark photon dark matter with GigaBREAD

The GigaBREAD collaboration presents the first broadband search for fixed-polarization dark photon dark matter in the 10.7–12.5 GHz range, utilizing a polarization-matched likelihood analysis to set kinetic mixing limits approximately 20% stronger than previous rescaled constraints while finding no evidence for such a signal.

Original author: Jialin Yu, Stefan Knirck, Gabe Hoshino, Rakshya Khatiwada, Alex Lapuente, David W. Miller, Andrew Sonnenschein

Published 2026-11-06

📖 4 min read🧠 Deep dive

CC BY 4.0

Original author: Jialin Yu, Stefan Knirck, Gabe Hoshino, Rakshya Khatiwada, Alex Lapuente, David W. Miller, Andrew Sonnenschein

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 author. For technical accuracy, refer to the original paper. Read full disclaimer

For decades, astronomers have known that the visible stars and galaxies make up only a tiny fraction of the universe. The rest is an invisible substance called dark matter, which holds galaxies together through its gravity but refuses to interact with light in any way we can easily detect. While we know it is there, we have no idea what it is made of. One leading theory suggests that dark matter might be composed of extremely light, invisible particles called dark photons. These particles would be similar to the photons that make up visible light, but they would be so faint and weak that they pass through ordinary matter without a trace. If they exist, they should be streaming through our solar system constantly, like a silent, invisible wind.

The challenge for scientists is that these particles are incredibly difficult to catch. To find them, researchers build sensitive instruments designed to convert a dark photon into a regular, detectable radio wave. This conversion is a rare event, and the signal is so faint that it is easily drowned out by the static of the universe. Furthermore, the nature of these particles might be more complex than previously thought. While many experiments assume the dark matter wind blows in random directions, a specific theory suggests that if these particles were created in a particular way in the early universe, they might all point in the same fixed direction, like a fleet of arrows frozen in flight. If this is true, the signal we detect would not be constant; it would rise and fall rhythmically as the Earth rotates, changing our angle relative to that invisible wind.

A team of physicists recently turned their attention to this possibility using a large, room-temperature experiment called GigaBREAD, located in the Chicago area. This device looks like a large, vertical metal cylinder with a cone in the center, designed to catch the faint radio waves that would be produced if a dark photon hit the metal surface. The experiment ran for several months, collecting tens of thousands of measurements of radio power across a specific range of frequencies, from 10.7 to 12.5 gigahertz. In previous analyses of this data, the researchers assumed the dark matter wind was random, meaning the signal strength would be the same regardless of the time of day. However, the team realized that if the dark photons have a fixed polarization, the signal should wiggle up and down over the course of a day as the Earth spins.

To test this, the researchers re-examined their old data with a new, more sophisticated method. Instead of looking for a constant hum, they looked for the specific pattern of a signal that changes strength in sync with the Earth's rotation. They built a statistical model that could track this daily wobble, accounting for how the detector's orientation changes relative to the fixed direction of the dark matter wind. They analyzed two separate runs of the experiment, one with the detector pointing straight up and another with it lying flat, to see if the pattern held up under different conditions.

The result was a clear null finding. The data showed no sign of the rhythmic modulation that would indicate a fixed-polarization dark photon signal. The radio waves measured by the detector were consistent with random background noise, with no hidden pattern emerging from the static. Because they found no evidence of the signal, the team used their results to set a new, stricter limit on how strong the interaction between dark photons and ordinary light could be. They determined that if these particles exist, their ability to convert into radio waves must be weaker than a specific threshold, roughly one part in a trillion.

This new limit is about twenty percent more sensitive than previous estimates that simply tried to guess how a fixed signal would look based on random assumptions. By using the actual timing of the Earth's rotation to guide their search, the researchers were able to squeeze more information out of the same data. While they did not find dark matter, they successfully ruled out a specific version of the theory where the particles have a fixed direction and are strong enough to be seen by this instrument. The work demonstrates that looking for these subtle, time-dependent changes in data can sharpen the search for the invisible universe, even when the particles themselves remain hidden.

Technical Summary: First Broadband Search for Fixed-Polarization Dark Photon Dark Matter with GigaBREAD

Problem Statement
Dark photon (DP) dark matter, if produced via a misalignment-like mechanism, may retain an approximately fixed polarization direction over cosmological and laboratory timescales. As the Earth rotates, the projection of this fixed polarization onto a detector's sensitivity axis varies with sidereal time, creating a time-dependent signal modulation. Previous broadband dish antenna searches, including the GigaBREAD experiment, have typically assumed that DP dark matter possesses a random polarization distribution. Under this assumption, the projection factor averages to $\langle \cos^2 \vartheta \rangle = 1/3$ over time, serving as a constant suppression factor in exclusion limits. However, this assumption may not reflect the true nature of the signal, potentially introducing biases in exclusion limit estimates and failing to utilize the time-dependent modulation information inherent in fixed-polarization scenarios.

Methodology
This work presents a reanalysis of GigaBREAD data acquired in the 44–52 $\mu$ eV mass range (10.7–12.5 GHz) using a polarization-matched likelihood framework. The analysis incorporates two datasets: a zenith-aligned run (GigaBREAD2023) and a horizontal-axis run (GigaBREAD2024).

Signal Model: The expected signal power $S_{kj}$ in the $j$ -th frequency bin of the $k$ -th spectrum is modeled as:
$S_{kj} = P_0(\chi, m_{DP}) \eta_{kj} C_j(m_{DP}) \cos^2 \vartheta(t_k)$
where $P_0$ is the conversion power dependent on the kinetic mixing parameter $\chi$ , $\eta_{kj}$ is the detection efficiency, $C_j$ is the frequency-bin response, and $\cos^2 \vartheta(t_k)$ represents the time-dependent projection of the fixed DP polarization vector $\hat{X}$ onto the detector's sensitivity axis $\hat{D}(t)$ . The modulation is expanded into sidereal harmonics (constant, daily, and twice-daily components) based on the detector's latitude and orientation.
Statistical Framework: A likelihood function is constructed to compare the measured residual power spectra against the signal model. To handle the unknown polarization angles ( $\theta_X, \phi_X$ ) as nuisance parameters, the authors employ both Bayesian and frequentist approaches:
- Bayesian: Marginalizes over the polarization angles using a uniform prior over solid angle.
- Frequentist: Profiles over the polarization angles by maximizing the likelihood.
  Due to the computational cost of evaluating the likelihood for $\sim 10^4$ spectra and millions of polarization hypotheses, the analysis utilizes a harmonic decomposition of the geometric factors. This allows the time-dependent sums to be precomputed, significantly accelerating the evaluation of the cross-correlation and template terms.
Significance and Limits:
- Discovery: Global Bayesian evidence ( $k_{global}^0$ ) and frequentist test statistics ( $q_0$ ) are computed across the mass range. Empirical null distributions are generated using 300 toy Monte Carlo datasets to calibrate global thresholds, accounting for the non-Wilks distribution caused by profiling over nuisance parameters.
- Upper Limits: In the absence of a signal, 90% confidence/credible upper limits on the conversion power $P_0$ and the kinetic mixing parameter $\chi$ are derived. The frequentist limits are determined via a Neyman construction calibrated against the "worst-case" polarization direction to ensure coverage.

Key Contributions

First Fixed-Polarization Search: This is the first broadband search specifically targeting fixed-polarization dark photon dark matter, moving beyond the standard random-polarization assumption.
Improved Sensitivity: By utilizing the time-dependent modulation information rather than a simple time-averaged suppression factor, the analysis achieves tighter constraints. The resulting limits on $\chi$ are approximately 20% lower (better) than a previous GigaBREAD reinterpretation that assumed fixed polarization but ignored modulation.
Statistical Framework: The paper demonstrates a robust method for handling unknown polarization directions in broadband searches using harmonic decomposition to manage computational complexity, applicable to both Bayesian and frequentist paradigms.
Detector Geometry Insights: The analysis identifies that for modulation-matched likelihoods, the optimal detector geometry (minimizing the worst-case polarization response) shifts from a half-opening angle of $54.7^\circ$ (for time-averaged analysis) to approximately $52.0^\circ$ ( $|\lambda| \approx 38.0^\circ$ ).

Results
The analysis finds no statistically significant evidence for a dark photon signal in the 44–52 $\mu$ eV mass range. Consequently, the authors set new upper limits on the kinetic mixing parameter $\chi$ .

The limits reach the level of $\chi \sim 10^{-12}$ .
These limits are comparable to the original GigaBREAD isotropic (random polarization) limits but represent a significant improvement over previous fixed-polarization reinterpretations that did not account for time-dependent modulation.
The combined analysis of the 2023 (zenith) and 2024 (horizontal) datasets provides constraints that are slightly stronger than the random-polarization assumption, benefiting from the complementary detector orientations.

Significance and Claims
The paper claims that incorporating polarization modulation and likelihood-based approaches significantly improves the sensitivity of broadband experiments to polarized dark photons. It demonstrates that broadband dish antennas, capable of observing the full bandwidth over extended periods, are particularly well-suited for this type of time-dependent analysis. The authors assert that their methodology can be directly applied to other broadband dark-photon searches and similar experiments involving time-dependent or polarization-dependent signals, offering a more rigorous framework for setting exclusion limits when the polarization state of dark matter is unknown but potentially fixed.

More like this